#arma3_tools | ARMA | Page 28

smoky halo Jan 4, 2019, 12:27 PM

#

_probabilityThold is a sum

#

so _probabilityThold[0] is probability[0] but _probabilityThold[1] is _probabilityThold[0] + probabiolity[1], _probabilityThold[2] is _probabilityThold[0] + _probabilityThold[1] + probaility[2]

#

etc

#

so the order doesnt matter

#

and surf is array of characters as in CfgSurfaceCharacters

rotund rampart Jan 4, 2019, 12:34 PM

#

I've updated tabler
https://github.com/bux/tabler/releases/tag/v0.9.0

enjoy 😃

GitHub

bux/tabler

Arma 3 Translation Helper. Contribute to bux/tabler development by creating an account on GitHub.

nocturne basin Jan 4, 2019, 12:35 PM

#

it still uses forms? urgh

rotund rampart Jan 4, 2019, 12:36 PM

#

it does ¯_(ツ)_/¯

obsidian sluice Jan 4, 2019, 12:36 PM

#

What's wrong with it? It just works 👀

rotund rampart Jan 4, 2019, 12:37 PM

#

no one likes Win Forms 😃

nocturne basin Jan 4, 2019, 12:40 PM

#

@obsidian sluice WinForms is just outdated AF

#

there is now WPF for desktop applications

obsidian sluice Jan 4, 2019, 12:40 PM

#

Oh, I was making fomr WF Arma launcher some time ago 😄

pearl beacon Jan 4, 2019, 12:40 PM

#

but they still work

nocturne basin Jan 4, 2019, 12:40 PM

#

or UWP if you are idiotic (as UWP essentially was useful but nowadays without actual windows mobile it is just some underdeveloped WPF)

obsidian sluice Jan 4, 2019, 12:40 PM

#

It had some cool listbox or something which WPF didn't have as default 😛

pearl beacon Jan 4, 2019, 12:40 PM

#

UI stuff is like all of JS

#

you can't follow

#

so can it even be outdated is the question!

nocturne basin Jan 4, 2019, 12:41 PM

#

@obsidian sluice well ... it is just different in WPF mostly 🤷 but i do agree, that some things (eg. the datagrid) are neater in forms

pearl beacon Jan 4, 2019, 12:42 PM

#

Qt or bust

#

works everywhere and is not from Microsoft 😛

rotund rampart Jan 4, 2019, 12:42 PM

#

I'd totally use WPF if I'd rewrite tabler ¯_(ツ)_/¯

pearl beacon Jan 4, 2019, 12:43 PM

#

I'd use Qt and make it work on Linux as well

rotund rampart Jan 4, 2019, 12:43 PM

#

electron 😛

pearl beacon Jan 4, 2019, 12:43 PM

#

yeah no

#

and then people wonder why RAM became so expensive, clearly related

obsidian sluice Jan 4, 2019, 12:44 PM

#

Just download more RAM dummy 4HEad

pearl beacon Jan 4, 2019, 12:44 PM

#

sideways to make it faster

nocturne basin Jan 4, 2019, 12:45 PM

#

@rotund rampart no ... just no to electron
if i would have time, i could link you to like a thousand links telling you why running applications in a browser container is a bad idea

obsidian sluice Jan 4, 2019, 12:45 PM

#

Yeah it doesn't sound good 😄

pearl beacon Jan 4, 2019, 12:45 PM

#

you need links for that? XD

obsidian sluice Jan 4, 2019, 12:45 PM

#

^ 😄

rotund rampart Jan 4, 2019, 12:45 PM

#

@nocturne basin i'm just teasing jonpas

nocturne basin Jan 4, 2019, 12:45 PM

#

WPF should also arive soon on linux btw

pearl beacon Jan 4, 2019, 12:46 PM

#

they said it won't

nocturne basin Jan 4, 2019, 12:46 PM

#

https://blogs.windows.com/buildingapps/2018/12/04/announcing-open-source-of-wpf-windows-forms-and-winui-at-microsoft-connect-2018/

pearl beacon Jan 4, 2019, 12:46 PM

#

open-source yes

#

and .NET Core whatever version

#

but not WPF

#

just go to their GitHub WPF project

#

and check issues 😛

#

https://github.com/dotnet/wpf/issues/48#issuecomment-444316938

#

there

#

wait

#

I meant to link this https://github.com/dotnet/wpf/issues/48#issuecomment-444198305

#

instead I linked the drama post, oh well, enjoy

nocturne basin Jan 4, 2019, 12:50 PM

#

it is not really that much of a problem
as i already said "soon" in a way that ment "the moment somebody ports this to OpenGL (or vulkan)"

smoky halo Jan 4, 2019, 12:50 PM

#

@vague shard the sum of probabilities should be <= 1. For example [0.5,0.5,0.1] will result in first 2 been chosen 50/50 and third one never chosen. [0.5] will result in first one chosen 50 % and default clutter chosen 50%

#

`probability[] = {0.98,0.03,0.02,0.015};
the way the community understood it so far:

sum may not be >=1
its a probability for each clutter
where above source suggests it checks a position for each probability one by one, and breaks if the random value is below probability
which also implies the values would need to be ascending, or would never be true`

sum cannot be > 1 , sum ==1 is ok
wrong. no particular order is required
if sum is < 1 means default clutter will be chosen as well, default as first clutter defined in >> "Clutter"

nocturne basin Jan 4, 2019, 12:59 PM

#

@rotund rampart not sure if i do repeat myself (could be that i forgot to mention) but, feel free to pick up what i did and continue on working inside of it 🤷

#

the StringtableEditor thingy i created and linked to here, is far from finished but written in WPF

#

and already contains the required basics

dawn palm Jan 4, 2019, 1:09 PM

#

excellent research @smoky halo
it now makes perfect sense why all entries after the 1.0 threshold has been reached, are ignored

bushlurker was right all along.

smoky halo Jan 4, 2019, 1:21 PM

#

Well it would take an extra line of code to normalise sum to fit it in 0..1 range

#

But this would remove fun out of arma config

dawn palm Jan 4, 2019, 1:27 PM

#

true again, but even the solution as-is is more elegant than i thought.

rotund rampart Jan 4, 2019, 1:29 PM

#

@nocturne basin I'll take a look at it

#

but time is short

nocturne basin Jan 4, 2019, 1:29 PM

#

https://github.com/X39/StringtableEditor everybodies time is limited @rotund rampart

#

but you already took the time to update, so you may want to spend those other hours to upgrade

smoky halo Jan 4, 2019, 1:39 PM

#

@dawn palm yes this 100 y.o. game continues to surprise 😂

orchid shadow Jan 4, 2019, 5:28 PM

#

Honestly what microsoft tried to do with WPF was nice but kinda was unwiedly and shit

#

There is a reason why most games run a fuking browser to render thier UI nowdays, just easier to work with.
Getting things to work nicely in WPF is a awful sometimes

#

The 3.0> of swifty is written in WPF, took longer then rewrtiing the whole backend :P

#

also everyone should check out Swifty 3, it's a lot better.

pearl beacon Jan 4, 2019, 5:40 PM

#

SwiftPbo is also used in DokanPbo 😄

nocturne basin Jan 4, 2019, 10:58 PM

#

Is there any valid critic on wpf besides €had no idea what I was doing so took ages" @orchid shadow ?

orchid shadow Jan 4, 2019, 11:10 PM

#

@nocturne basin Heh trying to bait me into a worthless dicussion with you.

#

No thanks sir.

#

Go back to your lair

nocturne basin Jan 4, 2019, 11:12 PM

#

There is no actual discussion present here
But barking "it is shit because It took me long to do it" is no valid critics

orchid shadow Jan 4, 2019, 11:13 PM

#

Yes, thats exactly what i said or was it?

#

Fucking read what i wrote.

nocturne basin Jan 5, 2019, 12:01 AM

#

is no valid critics take that advice for yourself

sacred flume Jan 5, 2019, 1:19 AM

#

I've been ok with WPF so far, but I'm not trying to do anything too complex with it

vague shard Jan 5, 2019, 10:10 AM

#

@smoky halo thanks for the research and helping understanding it properly 👍

orchid shadow Jan 5, 2019, 12:26 PM

#

@nocturne basin 😕

glossy inlet Jan 10, 2019, 11:47 AM

#

Moving here from #arma3_scripting as this channel makes more sense.
Thinking about making a Database extension using Intercept.
These are my ideas currently:

DB_Connection = db_createConnection [ip, port, username, password, database];
//or db_createConnection "configName" not sure if seperate config file or Arma config, probably seperate


DB_Cache_InsertQuery = db_prepareQuery 'INSERT INTO positions (name, x, y, z) VALUES (?,?,?,?)';

private _insertQuery = db_copyQuery DB_Cache_InsertQuery; //Such that you can prepare queries that you often reuse, maybe even with values already bound to the query

_insertQuery db_bindValue (name player); //Binds value to next available placeholder
_insertQuery db_bindValueArray (position player); //foreach value {Binds value to next available placeholder}. So this sets x, y and z


private _asyncResult = DB_Connection db_queryAsync _insertQuery;

_asyncResult db_bindCallback [{
    params ["_result", "_arguments"];
    private _affectedRows = db_resultAffectedRows _result;
    systemChat format ["The query has completed! %1 rows affected", _affectedRows];
    
    private _result1 = db_resultGetColumn 0; //_result is DBValue type.
    db_valueGetType _result1; //"ARRAY"
    private _resultRowArray = db_valueToArray _result1; //Converts DBValue which is array of more DBValue's to a SQF array out of string/number/arrays
    systemChat ("row1: " + str _resultRowArray);
}, argument1];

Thoughts and ideas? bindCallback will act like a eventhandler.

sick verge Jan 10, 2019, 11:52 AM

#

what for? What is the benefit of doing this in Intercept?

#

There already are extensions for database handling 🤔

smoky halo Jan 10, 2019, 11:52 AM

#

You should limit database comunication to calling stored procedures and be done with it

glossy inlet Jan 10, 2019, 11:52 AM

#

No need to plug everything into strings. Async callbacks. No need to compile/parseSimpleArray the returned values

smoky halo Jan 10, 2019, 11:53 AM

#

Anything a user would want to do can be made into a stored procedure like you make functions for sqf

sick verge Jan 10, 2019, 11:53 AM

#

💡

glossy inlet Jan 10, 2019, 11:54 AM

#

Yeah stored procedures are good for bigger projects. But for something where you just wanna quickly add some DB stuff.
If you want stored procedures you can just do a CALL query. Then you also control the security of all this on your db server, you can disallow everything besides CALL for example

nocturne basin Jan 10, 2019, 11:57 AM

#

- DBCONNECTION
- DBSET
- DBCMD

Commands:
- createConnection STRING -> DBCONNECTION
- DBCONNECTION createCommand STRING -> DBCMD
- DBCMD addParameter [Name:STRING, Content:ANY] -> NIL
- execute DBCMD -> SCALAR (Affected records)
- executeQuery -> DBSET
- isReady DBSET -> BOOL
- next DBSET -> NIL
- headers DBSET -> ARRAY (strings)
- row DBSET -> ARRAY```

#

and again: there is reasons for instead of having a single row DBSET command to have a DBSET getValue Index:SCALAR -> ANY command

glossy inlet Jan 10, 2019, 12:00 PM

#

Don't know if I really need that stuff. Could just turn the whole result into a [[col,col,col], [col,col,col]] array and let the user do the rest in SQF if he wants.
Performance wise ofc not a good idea to deserialize everything to SQF if you don't need everything. But if you don't want to have all the results of your query, why did you query for them then :u

nocturne basin Jan 10, 2019, 12:01 PM

#

getValue over row is actually just a size concern + the advantage of being able to use DBSET getValue "column"

glossy inlet Jan 10, 2019, 12:01 PM

#

oh right. Didn't think about retrieving values by colum name

smoky halo Jan 10, 2019, 12:01 PM

#

if you allow multi result quires, the array returned is quite deep

nocturne basin Jan 10, 2019, 12:01 PM

#

turning a whole result into a gigantic array is stupid for many reasons i do not have to explain i guess 😛

#

it cannot be @smoky halo

#

select always returns a single table

#

in all dbs

smoky halo Jan 10, 2019, 12:02 PM

#

you can have multiresult

nocturne basin Jan 10, 2019, 12:02 PM

#

--> A.Column, B.Column
still single table

smoky halo Jan 10, 2019, 12:02 PM

#

query;query;...

nocturne basin Jan 10, 2019, 12:02 PM

#

that again would be against the command idea

glossy inlet Jan 10, 2019, 12:02 PM

#

jup. I guess then _result would be an array of result sets

nocturne basin Jan 10, 2019, 12:03 PM

#

just avoid it at all

#

as alternative: can be done with the above commands too

#

but then at some point headers would change

#

there simply never is nested results
just multiple you receive

glossy inlet Jan 10, 2019, 12:04 PM

#

only nesting beyond the row->column would be if there is a array datatype in the column. like position

smoky halo Jan 10, 2019, 12:06 PM

#

@nocturne basin you dont want to go to the database for every single value, you often want to get user shit out and it could contain many items as well as you might want to do that for several users at once.

nocturne basin Jan 10, 2019, 12:35 PM

#

@smoky halo if you fucked up your queries that is your fault

smoky halo Jan 10, 2019, 12:37 PM

#

what about catering for idiots philosophy?

nocturne basin Jan 10, 2019, 12:51 PM

#

idiots would not do something like query "select * from users; select * from whatever"

#

and even if, you could provide them with, as i said, just a next method that is suddenly changing the actual result set

glossy inlet Jan 10, 2019, 12:52 PM

#

Man.. building armake is made hard by you having to get the openssl library.
It doesn't have to be hard at all though. Add a git submodule for wolfssl, link in cmake project, boom everything is handld automatically...

nocturne basin Jan 10, 2019, 12:52 PM

#

or concat the headers for them (depending on the DB used)

scenic canopy Jan 10, 2019, 1:01 PM

#

@nocturne basin you forgot adding some params without escaping

nocturne basin Jan 10, 2019, 1:03 PM

#

no i did not @scenic canopy

#

no need for remote execution at any point in any software

scenic canopy Jan 10, 2019, 1:04 PM

#

query format ["select * from users; select * from whatever where id = %1", _number]

#

then you don't need stored procedures or prepared statements!!

nocturne basin Jan 10, 2019, 1:05 PM

#

what exactly you wanted to show me with that?

glossy inlet Jan 10, 2019, 1:05 PM

#

Took me just an hour to get the project set up. Mysql c++ connector with wolfssl and intercept. All in a cmake project.
And it just successfully built into a single statically linked dll. (after adding a include that's missing in the mysql connector library, don't ask me why)
If only the build experience would always be that nice...
Now I guess I only have to find out if t works 😄

nocturne basin Jan 10, 2019, 1:06 PM

#

if you have, i would gladly join your efforts

scenic canopy Jan 10, 2019, 1:06 PM

#

@glossy inlet you can add the release download link to cmake and avoid the git submodule

karmic niche Jan 10, 2019, 1:07 PM

#

@scenic canopy i also don't understand the point you're trying to make with that sql

#

Are you saying that you should be able to format like above???

scenic canopy Jan 10, 2019, 1:08 PM

#

if I add a troll face, would it be better? 😛

smoky halo Jan 10, 2019, 1:08 PM

#

c++ connector looks like Java port, 🤢

karmic niche Jan 10, 2019, 1:08 PM

#

Much better! :D

nocturne basin Jan 10, 2019, 1:09 PM

#

@smoky halo the JDBC is actually some extremly good design for generic DB access

glossy inlet Jan 10, 2019, 1:09 PM

#

You could use something like that format for a more "convenient" bindValue thing. But I don't really like that.
I mean something like queryWithBoundValues = db_formatQuery [SourceQuery, value1, value2, value3] ugh

nocturne basin Jan 10, 2019, 1:09 PM

#

no

#

just no

#

stick with what i gave ya

#

just ... instead of ? use ?name

#

and then people can do DBCMD addParameter [name, value]

#

or with just ? only DBCMD addParameter value

karmic niche Jan 10, 2019, 1:11 PM

#

Call a function once for each parameter? Why?

scenic canopy Jan 10, 2019, 1:11 PM

#

@glossy inlet https://cmake.org/cmake/help/latest/module/ExternalProject.html#examples

nocturne basin Jan 10, 2019, 1:11 PM

#

because named parameters are a thing @karmic niche

#

and thats how you do it in frameworks

karmic niche Jan 10, 2019, 1:12 PM

#

In Frameworks, you can use several named parameters at once

nocturne basin Jan 10, 2019, 1:12 PM

#

uhm ... yup? that is why i propose for a addParameter

karmic niche Jan 10, 2019, 1:13 PM

#

Calling the function once for each parameter would be tedious

glossy inlet Jan 10, 2019, 1:13 PM

#

Don't really see the need for named parameters. Also hard to implement with the database connector library.

karmic niche Jan 10, 2019, 1:13 PM

#

setParameters, if anything imo

nocturne basin Jan 10, 2019, 1:13 PM

#

<14:13:36 SimplifySoft.Pecuniarius.Server.Database.PecuniariusContext> (#42446020)
<14:13:36 SimplifySoft.Pecuniarius.Server.Database.PecuniariusContext> (#42446020)      -- p_0: '1/10/2019 2:13:36 PM' (Type = DateTime, IsNullable = false)
<14:13:36 SimplifySoft.Pecuniarius.Server.Database.PecuniariusContext> (#42446020)      -- p_1: '{
  "SimplifySoft.Pecuniarius.Data.Net.AccountingRepository.Receipt": {
    "TimeStampCreated": "2019-01-10T14:13:36.4133742+01:00",
    "TimeStampLastChanged": "2019-01-10T14:13:36.4133742+01:00",
    "Entries": []
  }
}' (Type = Object, IsNullable = false)```

#

inb4

#

that is how postgres receives values

#

mysql won't be any different

#

that is why the design in many top level interfaces is addParameter not addParameters

karmic niche Jan 10, 2019, 1:14 PM

#

Disclaimer: I'm spoiled by ORMs and can't stand regular SQL

dusky dune Jan 10, 2019, 1:14 PM

#

why not both?

nocturne basin Jan 10, 2019, 1:14 PM

#

same deal there @karmic niche 😜

#

just stop using the implementation APIs and rather use the top-level interfaces

dusky dune Jan 10, 2019, 1:15 PM

#

i dont know shit about sql OR ORMS (srry stack), but using qgis features or so you wanna add attributes to shapefiles? You can either add a single, or just use addAttributes in a list straight away

nocturne basin Jan 10, 2019, 1:16 PM

#

both would require array checking

connection whatevercommand [["key", "value"]]
// or
connection whatevercommand ["value1", "value2"]```

glossy inlet Jan 10, 2019, 1:16 PM

#

Okey take it back. Named parameters are not hard.
How about

db_bindNamedValue ["name", value]
db_bindNamedValueArray [["name", value], ["name", value]]

dusky dune Jan 10, 2019, 1:16 PM

#

👌🏻

nocturne basin Jan 10, 2019, 1:16 PM

#

do not forget to check each entry

#

personally, i would still prefer the addParameter syntax simply because it will look cleaner in code

glossy inlet Jan 10, 2019, 1:19 PM

#

Edited my above post. If there is an array variant there ofc is one without array too. That's essentially the same as yours

nocturne basin Jan 10, 2019, 1:19 PM

#

can live with that 🤷

#

though ... maybe combining both would be better 🤔

#

ohh ... and i hate your naming 😄

glossy inlet Jan 10, 2019, 1:25 PM

#

I don't want to conflict with anything that might exist elsewhere. So db_ prefix it is

nocturne basin Jan 10, 2019, 1:26 PM

#

was not talking about the prefix

glossy inlet Jan 10, 2019, 1:27 PM

#

Well in normal mysql connector it's called bind, so I use bind too.
Also it's called a database query, so I use query too instead of command

#

No need to make up new naming if naming already exists

nocturne basin Jan 10, 2019, 1:27 PM

#

mysql is braindead

better to use actual database interface names then the ones mysql, postgres or whatever one wants to use

glossy inlet Jan 10, 2019, 1:35 PM

#

Oh nice. {params ["_row"]; ...} forEach _resultSet So many ideas 😄

nocturne basin Jan 10, 2019, 1:36 PM

#

nah ... do not do that @glossy inlet

#

bad idea

#

or only in spawn

#

so you can suspend the script until the next result is ready

glossy inlet Jan 10, 2019, 1:37 PM

#

Good idea too. But so far I've only written unscheduled forEach loops in Intercept. Scheduled means I need special callstack stuff

nocturne basin Jan 10, 2019, 1:38 PM

#

yup ... and if you want to implement something like that for something that requires waiting seconds in worst case, you want to avoid at all cost that some idiot is comming along (or just by accident) and uses it inside of a forEach loop

#

if they want to make that mistake ... one can use a while loop with waituntil or whatever

glossy inlet Jan 10, 2019, 1:39 PM

#

Also nice for async results. Implement a sleep like thing that just waits till result is done in scheduled.
VS a waitUntil loop calling the extension every half second like it's done now with extdb

nocturne basin Jan 10, 2019, 1:40 PM

#

can be done using next and hasNext

#

next would then wait, hasNext just checks

glossy inlet Jan 10, 2019, 1:40 PM

#

I could hide it inside the forEach loop. Would be easier to use.

nocturne basin Jan 10, 2019, 1:41 PM

#

🤷

#

but requires you to actually halt the VM until it is available

#

the way i imagined it was no actual waiting

#

but rather all just work on existing data

#

this way, you have less trouble and can just get it done

#

doing the fancy things later

glossy inlet Jan 10, 2019, 1:42 PM

#

I prefer doing the waiting on engine level. Rather than the user having a loop that waits for him

nocturne basin Jan 10, 2019, 1:42 PM

#

doing the fancy things later

#

do what you have to 😉 i might be one of the first users when it is done

nocturne basin Jan 10, 2019, 2:07 PM

#

ye ... and i now left the DayZ Modders Discord
still gonna make that enscript VM at some point probably though

#

(as long as somebody can provide me with what i need for that 🙈)

karmic niche Jan 10, 2019, 2:20 PM

#

Why exactly did you leave?

nocturne basin Jan 10, 2019, 2:26 PM

#

Was not granted the access to the internal channels
Also... They are not really helpful there for what I planned on doing

karmic niche Jan 10, 2019, 2:27 PM

#

Have you submitted something through the google form?

#

You could easily get access to the tools role, IMO

nocturne basin Jan 10, 2019, 2:27 PM

#

"check leaked Code" was the answer to the question wether somebody could hint me at interna of enscript

#

Yes, did so

karmic niche Jan 10, 2019, 2:28 PM

#

That's probably because "we actually don't know how the language works, yet" isn't something they felt like saying 😄

#

Especially that that discussion was in september

glossy inlet Jan 10, 2019, 2:30 PM

#

There are very few people in there that operate on the level that you need to get that info

karmic niche Jan 10, 2019, 2:30 PM

#

btw. you reminded me that I was planning to apply myself, there

#

So yeah, what Dedmen said. Also, I think they didn't fully understand the extent of the work that you wanted to do.
And checking the leaked code is um... problematic. On one hand, it will give you the most accurate answer. On the other hand, it's not exactly ( 😉 ) legal, and is certainly out of date at this moment

nocturne basin Jan 10, 2019, 2:47 PM

#

Applied the day that Annoncement Was made (maybe 10 Minutes later)
Did not received anything yet

I explained everything multiple times... Answers always have been the same "no need for that" "check leaked source" etc.

#

And I am Well aware that there are only a few people capable of telling me what I need to know

dusky dune Jan 10, 2019, 3:50 PM

#

not entirely sure how much of a good idea is it to have an important discord like that run to by community members either

#

inb4 political drama surrounding mods 👌🏻

glossy inlet Jan 10, 2019, 3:51 PM

#

DayZ discord exists too. And it also has modding channels..

#

They probably are not there because illegal stuff that violates EULA get's banned on sight there..
In their own Discord in their private channels they can hack and violate EULA all they want

#

That was easy..

DB_Connection = db_createConnection [ip, port, username, password, database];
_insertQuery = db_prepareQuery 'INSERT INTO positions (name, x, y, z) VALUES (?,?,?,?)';
_insertQuery db_bindValue (name player); //Binds value to next available placeholder
_insertQuery db_bindValueArray (position player); //foreach value {Binds value to next available placeholder}. So this sets x, y and z
_result = DB_Connection db_query _insertQuery;

systemChat (str db_resultToArray _result);

already all implemented. Should be enough to test. I might test when I get home in half an hour. Just missed my bus just to get this released first
I pushed a test build https://github.com/dedmen/Intercept_database/releases It is called sqf-assembly because I don't have a pbo packer to hand right now

No async stuff yet, that'll require quite a bit more work

scenic canopy Jan 10, 2019, 4:01 PM

#

thanks for inspiring me to get started with intercept plugins again 😄

glossy inlet Jan 10, 2019, 4:02 PM

#

exactly 3 hours of work, from 0 to a working Intercept based database extension with CMake build :u
That almost went too swiftly

smoky halo Jan 10, 2019, 4:48 PM

#

Do you use dynamic linking?

glossy inlet Jan 10, 2019, 4:49 PM

#

no

#

I would need to put other libraries into arma directory.... Or fix the Intercept workaround that I should've implemented months ago.

smoky halo Jan 10, 2019, 4:50 PM

#

So you static linked c++ connector?

glossy inlet Jan 10, 2019, 4:50 PM

#

yeah

smoky halo Jan 10, 2019, 4:50 PM

#

Ok

glossy inlet Jan 10, 2019, 4:50 PM

#

Well. I just set BUILD_STATIC to true in cmake... 😄

#

I forgot to register db_prepareQuery 🤦

glossy inlet Jan 10, 2019, 6:08 PM

#

ah ffs. There is a new very nice database interface that's not JDBC, but that only works with mysqlx which is some oracle specific crap :u
Switching to the "legacy" JDBC interface requires me adding boost -.- but atleast I get easier async

fallen stone Jan 10, 2019, 7:14 PM

#

There is nothing wrong with boost

nocturne basin Jan 10, 2019, 7:15 PM

#

size

#

compiletime

#

size

#

size

#

compiletime

#

think i got everything wrong with it 🤔

fallen stone Jan 10, 2019, 7:16 PM

#

You don't need the entirety of boost

#

Compile times aren't ideal, but they are a bucket load of headers

glossy inlet Jan 10, 2019, 7:17 PM

#

Takes like 10 minutes to clone the git submodule. I'd have to make my own git repo for the exact subset that I need

fallen stone Jan 10, 2019, 7:17 PM

#

Which repo?

glossy inlet Jan 10, 2019, 7:17 PM

#

Will do that someday. boost-cmake

nocturne basin Jan 10, 2019, 7:17 PM

#

last time i tried to use anything from boost, the "not all needed" thing needed more and more subsets

fallen stone Jan 10, 2019, 7:17 PM

#

boost-cmake repo..?

glossy inlet Jan 10, 2019, 7:18 PM

#

https://github.com/boost-cmake/boost this. Hoping it works out correctly with cmake

fallen stone Jan 10, 2019, 7:18 PM

#

Last time you mentioned that it wasn't true @nocturne basin

#

What exactly is that repo for?

nocturne basin Jan 10, 2019, 7:19 PM

#

https://discordapp.com/channels/105462288051380224/105464579600977920/518472507519533057

fallen stone Jan 10, 2019, 7:19 PM

#

https://canary.discordapp.com/channels/105462288051380224/105464579600977920/518487375043624971

nocturne basin Jan 10, 2019, 7:19 PM

#

we already had the entire conversation

fallen stone Jan 10, 2019, 7:19 PM

#

Yeah

#

Exactly

nocturne basin Jan 10, 2019, 7:20 PM

#

and asio required more then "just" that to get the examples working or ... itself

smoky halo Jan 10, 2019, 7:20 PM

#

I just skip any answer on stack overflow mentioning boost

fallen stone Jan 10, 2019, 7:20 PM

#

Why?

smoky halo Jan 10, 2019, 7:21 PM

#

Because I like shit lean

fallen stone Jan 10, 2019, 7:21 PM

#

That makes no sense

#

You're talking about header files

smoky halo Jan 10, 2019, 7:21 PM

#

It does to me

#

Not talking about size

fallen stone Jan 10, 2019, 7:22 PM

#

What's not lean after your compilation has eaten it?

#

Or do you prefer to use the newer standards, which are just ported boost modules 🙃

nocturne basin Jan 10, 2019, 7:24 PM

#

https://stackoverflow.com/questions/569198/boost-considered-harmful

#

boost is just a gigantic pile of neat code
but utterly useless unless you have a copy already locally ready of it in entirety

fallen stone Jan 10, 2019, 7:24 PM

#

Did you read the answers?

nocturne basin Jan 10, 2019, 7:24 PM

#

especially if you just need a simple subset

#

i did post it for the actual reasons in the top post

smoky halo Jan 10, 2019, 7:25 PM

#

C++ is already full of goodies will take ages to explore stuff they added just with c++17 don’t want to learn a 3rd party lib on top

fallen stone Jan 10, 2019, 7:25 PM

#

You can blame the state of C++ for that, unfortunately

nocturne basin Jan 10, 2019, 7:25 PM

#

it is a sorta meta post anyways

fallen stone Jan 10, 2019, 7:25 PM

#

Most of the features you're going to learn in C++XX are simply from boost

#

As will continue

nocturne basin Jan 10, 2019, 7:26 PM

#

main difference: i do not need to download boost in its entirety to use those features afterwards

#

which is the whole point

smoky halo Jan 10, 2019, 7:26 PM

#

Yeah don’t mind that as long as it is part of std

nocturne basin Jan 10, 2019, 7:27 PM

#

and the reason dedmen said "at some point i may create a subset repo of boost cmake"

fallen stone Jan 10, 2019, 7:27 PM

#

No instead, you faff about with gcc for ages 😄 ?

#

You don't need all of boost, especially if you're using a build gen like CMake

smoky halo Jan 10, 2019, 7:27 PM

#

What do you mean?

nocturne basin Jan 10, 2019, 7:28 PM

#

you will still need to download the whole boost suite ... regardless of what you do ...

#

in the end, you include a small subset, yes

fallen stone Jan 10, 2019, 7:28 PM

#

No, you do not

nocturne basin Jan 10, 2019, 7:28 PM

#

but to get that subset working, you need to download all bloody whooping crap

fallen stone Jan 10, 2019, 7:28 PM

#

Depends entirely on the deps

smoky halo Jan 10, 2019, 7:29 PM

#

No I mean what is faff about?

fallen stone Jan 10, 2019, 7:29 PM

#

hmm?

smoky halo Jan 10, 2019, 7:30 PM

#

No instead, you faff about with gcc for ages 😄 ?

fallen stone Jan 10, 2019, 7:30 PM

#

I seem to remember X39 having a fist fight with compiling a later version of gcc

smoky halo Jan 10, 2019, 7:31 PM

#

Oh I see it was for X39

nocturne basin Jan 10, 2019, 7:31 PM

#

mostly because it was the same problem as with asio

#

after trying to compile the first module, another baffled me

#

going on and on

#

and in the end ... a simple old gcc installation caused troubles

#

which to find took the longest frking time -.-

#

to realize, to uninstall fully etc.

#

because hur dur long story and many dependencies

fallen stone Jan 10, 2019, 7:36 PM

#

Might also be worth noting, keeping an installation of boost around, means you don't have to mess around. Don't have to download it every time you use it

#

If you're using linux, then the world is your oyster, and a stress free one

#

If you're using VS, take a look at vcpkg, worked well for me last time

nocturne basin Jan 10, 2019, 7:37 PM

#

Might also be worth noting, keeping an installation of boost around, means you don't have to mess around.
which defeats its usage pretty much entirely for github

#

hell even those fancy projects utilizing test suites etc. are stupid for github simply because they contain a gigantic set of tests in their main repo which, most of the time, is 4x the size of the sources

fallen stone Jan 10, 2019, 7:39 PM

#

I'm not sure what you mean, boost is a dependency

#

You don't need to run the test suites either?

#

Upload it all if it floats your boat

glossy inlet Jan 10, 2019, 7:51 PM

#

tried to let cmake auto-install boost.. Ran out of disk space 😄

fallen stone Jan 10, 2019, 7:56 PM

#

😄

meager falcon Jan 11, 2019, 1:50 PM

#

@glossy inlet why not use some key value based database layout. Fits Arma way better actually. I rewrote the Epoch database extension which originally used redis. Made it so that it could also use MySQL

Planned on integrating SQLite and maybe another pure key value db option

glossy inlet Jan 11, 2019, 1:51 PM

#

I already have key-value hashmaps in intercept_cba
key-value is not really efficient though is it?

meager falcon Jan 11, 2019, 1:52 PM

#

In case of redis it actually is. Also when implemented well, MySQL can keep up quite well too

#

you only need a single table with 2 fields

#

Maybe 3 if you want TTL or something like that

glossy inlet Jan 11, 2019, 1:55 PM

#

I don't know what other people need. But I feel like key-value is a little to cumbersome if you want anything relational

#

People can just implement key-value stuff with my extension if they want to ¯_(ツ)_/¯

meager falcon Jan 11, 2019, 1:56 PM

#

Actually I don't know any use case in Arma scripts which actually needs a select query that is way different from 'where x = id'

glossy inlet Jan 11, 2019, 1:56 PM

#

performance optimization, only spawn vehicles if player get's in range

#

life server messaging system. Find all messags from player1 to me

meager falcon Jan 11, 2019, 1:57 PM

#

Can be done with the key name actually ^^

glossy inlet Jan 11, 2019, 1:58 PM

#

As I said. If people want they can implement key-value with my extension

#

I don't know why I should limit it so that it becomes unusable for anyone who wants something else

#

Also key -value would need to be strings, Would be inefficient if you'd store numbers like object positions

meager falcon Jan 11, 2019, 4:18 PM

#

True that

#

Just wanted mention it

#

How about actual query commands

#

Like dbUpdate, dbInsert

#

BTW, is somebody interested in a general purpose Arma Dockerimage?

#

If so, I need someone to test it 😅

smoky halo Jan 11, 2019, 4:23 PM

#

What’s that?

glossy inlet Jan 11, 2019, 4:24 PM

#

Don't see what use DBInsert would be. What would the difference be to a normal query?

meager falcon Jan 11, 2019, 4:27 PM

#

Ease of use for binding

glossy inlet Jan 11, 2019, 4:27 PM

#

But.. What is the difference. I don't see any difference vs normal query?
normal queries are already easy to bind. Don't see what I would do differently with a seperate insert command

#

Except maybe that I return just lastInsertID instead of a result type. but that's just one line of code saved.

meager falcon Jan 11, 2019, 4:29 PM

#

@smoky halo Just Google for docker, basically a kind of virtualization

It's about creating an isolated environment for the Gameserver

#

Well.. I just think that people would create sqf function wrappers for all the queries

glossy inlet Jan 11, 2019, 4:31 PM

#

yeah.. But... Still. What difference would a seperate insert command make?

#

In what way would it be any different than a normal query

meager falcon Jan 11, 2019, 4:31 PM

#

Saving some lines of code.. nothing other than that

glossy inlet Jan 11, 2019, 4:31 PM

#

Which line? I don't see any that would be saved

#

you still have to write the insert query, just same as with the db_query command

#

That's both.. 1 line of code

#

_query = prepareQuery "INSERT INTO..."
db_executeQuery _query

vs

_query = prepareQuery "INSERT INTO..."
db_insert _query

Where is the difference?

meager falcon Jan 11, 2019, 4:33 PM

#

I think of something like this: connection dbInsert [tablename, [values]]

glossy inlet Jan 11, 2019, 4:34 PM

#

ah. Well.

_query = prepareQuery "INSERT INTO tablename .."
_query bindValueArray _values;
db_executeQuery _query

vs

dbInsert [tablename, [values]]

? Meh.

meager falcon Jan 11, 2019, 4:34 PM

#

Scripters usually want as much abstraction as possible

nocturne basin Jan 11, 2019, 4:34 PM

#

lol

#

🤣

#

either you can bloody use a database properly or you can't

meager falcon Jan 11, 2019, 4:35 PM

#

Well, "sqf scripters" at least 😅

nocturne basin Jan 11, 2019, 4:35 PM

#

the level of abstraction you expect is impossible to archive without numerous checks

#

a dbInsert command in the way as you propose it, would not be possible unless one checks how many columns are available

glossy inlet Jan 11, 2019, 4:36 PM

#

And then the user also always has to provide all columns. leaving something null is not possible.. Okey I guess you can insert nil's.

nocturne basin Jan 11, 2019, 4:36 PM

#

you actually have to allow nil for null

meager falcon Jan 11, 2019, 4:36 PM

#

Nope. The columns just need default values

#

And even if you write sqf queries, you would need to do that to make it fail save

glossy inlet Jan 11, 2019, 4:37 PM

#

But that hurts readability alot. With inesrt query you have the column names right there and know what's what.
Your method would break when someone adds a column. And the scripter wouldn't see it. He also doesn't see if he mixes up two values

meager falcon Jan 11, 2019, 4:38 PM

#

I see your point there tho

nocturne basin Jan 11, 2019, 4:39 PM

#

assuming the following:

User
-------------
Id: Scalar (default: DBGenerated on Insert)
Created: String (default: DBGenerated on Insert)
Username: String (default: unset)
Money: Scalar (default: 200)

how would your "idea" look like?

#

what would make things simpler

meager falcon Jan 11, 2019, 4:42 PM

#

Optionally you could also have an array for column names and one for the matching values

glossy inlet Jan 11, 2019, 4:42 PM

#

_query = db_prepareQuery "INSERT INTO User (Username, Money) (?,?)"
_query db_bindValueArray [name player, 0];
_connection db_execute _query;

vs
_connection db_insert ["User", [nil,nil, name player, 0]]
Oh wait.. Now Id and Created are nil. Oof.

#

I have the feeling it's more overcomplicating things rather than making it easier.
Yeah the users don't have to write SQL anymore with your method. But they still have to know how to set up the database, so they probably also know how SQL works

meager falcon Jan 11, 2019, 4:45 PM

#

Mod contributers who can't write SQL could still use DB commands

#

you can integrate other DB types (non SQL) more easily

#

I don't dislike the way you currently use, just throwing in some ideas

karmic niche Jan 11, 2019, 4:50 PM

#

_connection db_insert ["User", [["Username", name player], ["Money", 0]]]

#

You can also do it the "bind stuff" way, if you want to have the actual arguments be in a separate line of code

#

having db_select, db_insert, db_update and db_delete would make much sense IMO

#

Sure, you can do everything with a db_query, but in the same way, you can do everything with PUSH xxx, CALL yyy commands in asm, instead of using higher-level languages and calling functions "normally" 😃

nocturne basin Jan 11, 2019, 4:54 PM

#

even though i do like that "syntax" i am still against it
simply because it is not really required at all

#

also, what should db_insert return?

#

all of those are just saving literally the "db_execute*" commands

karmic niche Jan 11, 2019, 4:55 PM

#

I can also see that being potentially a bit more bulletproof - prevents typos in "select", "insert" etc...

nocturne basin Jan 11, 2019, 4:56 PM

#

how would you want a db_select to look like? you cannot make this "bulletproof"
only thing this does is saving a single command + a few words

karmic niche Jan 11, 2019, 4:56 PM

#

also, what should db_insert return?
does it need to actually return anything other than success/failure?

nocturne basin Jan 11, 2019, 4:57 PM

#

as the whole show is techincally async

karmic niche Jan 11, 2019, 4:57 PM

#

saving a single command + a few words
And this is a bad thing? This is literally why people write wrappers. Why not do that for them, in that case?

glossy inlet Jan 11, 2019, 4:57 PM

#

some people might want last insert id

karmic niche Jan 11, 2019, 4:59 PM

#

Yeah, whatever, my point was that there is no point in making the library harder to use than it can potentially be, especially that it's not like anyone will spend hours implementing that feature.

nocturne basin Jan 11, 2019, 4:59 PM

#

because connection db_select ["table", ["field"]] is literally SELECT field FROM table

karmic niche Jan 11, 2019, 4:59 PM

#

I'd still use the former. I don't have time to spend my time debugging my typos

nocturne basin Jan 11, 2019, 5:00 PM

#

you still will spend that time

glossy inlet Jan 11, 2019, 5:00 PM

#

can still typo the table name or column names ^^

nocturne basin Jan 11, 2019, 5:00 PM

#

feild != field

#

which are the more common typos in DB queries

karmic niche Jan 11, 2019, 5:00 PM

#

No, guys... seriously?

glossy inlet Jan 11, 2019, 5:01 PM

#

Will implement them. But don't see the usefulness really

karmic niche Jan 11, 2019, 5:01 PM

#

A hundred of wrappers simplifying writing code for programmers by one single line of code each time is a hundred of lines saved in total (just please, don't take that literally...)

#

Less code written outside a library = less potential bugs, typos, mistakes. Yes, you can mistype the field name. Sure! But if I had the choice, I'd use the function that lets me make less mistakes, and does that for free...

nocturne basin Jan 11, 2019, 5:05 PM

#

we talk about queries
unless you can abstract away literally everything, there is no sense in abstracting away those few commands

#

with the exception being, if there would be multiple end points

#

that would use different languages

karmic niche Jan 11, 2019, 5:07 PM

#

Let's just say that I disagree and end the discussion here 😃

meager falcon Jan 11, 2019, 5:09 PM

#

Also, how about performance in this case.
Only one invocation instead of multiple

Sure the array creation has a cost, but its quite low in comparison I guess

glossy inlet Jan 11, 2019, 5:11 PM

#

better perf in SQF land yeah. Same work in extension. But I can move the work to seperate thread

glossy inlet Jan 11, 2019, 6:21 PM

#

It wurks! https://s.sqf.ovh/arma3_x64_2019-01-11_19-21-20.png
Download: https://github.com/dedmen/Intercept_database/releases/tag/0.2

nocturne basin Jan 11, 2019, 6:23 PM

#

nice

#

sooo ... lets start with the endless task of creating a proper life mission now i guess

glossy inlet Jan 11, 2019, 6:24 PM

#

Nice joke. Almost believed you

nocturne basin Jan 11, 2019, 6:24 PM

#

you better do
because i am 100% serious

#

fuck sake ... if UI stuff wasn't that ugly in arma ... even my XMedSys rewrite would have been done already

#

and in regards of the life mission: i got told "do it better" by those idiots multiple times

#

and no, i am not really a friend of people telling me i am too stupid to do something (which should not be a surprise to ya 😛)

#

though ... arma.studio rewrite still needs to be done

#

🤔

#

sooo much work

#

and enscript VM project not even started

#

ohhh boy
2019 will be a bloody busy project if i do not find some girl to give me an excuse for all of that

#

just to note: i actually started already back in 2017 https://imgur.com/KXNCzjE

#

but never got around doing my own db stuff as i hate extdb

glossy inlet Jan 11, 2019, 6:32 PM

#

Same. extdb is also quite cumbersome to set up with weird memory allocator stuff. I hope this thing will be same kind of plug-n-play on linux too

#

My Arma game today just dissolved.. So I guess I have time to implement async now... ugh.. I don't wanna...

meager falcon Jan 11, 2019, 6:57 PM

#

Back then I just accepted a the name of the function that would be called once the operation was done

#

Actual code as callback is kinda "meh" I guess

glossy inlet Jan 11, 2019, 7:00 PM

#

Well.. You can just plug a function name in there too

#

CODE type variable, or inline CODE doesn't make a difference

meager falcon Jan 11, 2019, 7:03 PM

#

I actually meant a string, I fetched the code from mission namespace

#

Otherwise you will have to pull the code all along to the end

glossy inlet Jan 11, 2019, 7:05 PM

#

Don't see why anyone would wanna do that

#

I just store a variable. It's not like that's any effort

meager falcon Jan 11, 2019, 7:07 PM

#

If it doesn't have a large performance impact, then that's better yeah 😅

glossy inlet Jan 11, 2019, 7:08 PM

#

just moving around a pointer and incrementing a refcounter ¯_(ツ)_/¯

#

storing a string would be the exact same actually

cinder meteor Jan 11, 2019, 7:31 PM

#

what is the 'async' thing you are going to do?

glossy inlet Jan 11, 2019, 7:33 PM

#

Callback once result is ready.
Or scheduled script suspension until result is ready. Like a waitUntil kinda

cinder meteor Jan 11, 2019, 7:34 PM

#

Database-related?

glossy inlet Jan 11, 2019, 7:34 PM

#

yes...

#

We were talking about the Database Extension the whole day.
So yep. It's database related ^^
Here is a example with a callback: https://discordapp.com/channels/105462288051380224/105464579600977920/532888283311767563

cinder meteor Jan 11, 2019, 7:37 PM

#

I see now, because with intercept you can call your SQF callback from the .dll

#

but you could pump callbacks from your .dll on the sqf side 🤷 I guess

glossy inlet Jan 11, 2019, 7:39 PM

#

what do you mean?

cinder meteor Jan 11, 2019, 7:40 PM

#

hmm... on each frame, ask the db extention for finished queries, then call the appropriate callback... didn't think much of it tbh

glossy inlet Jan 11, 2019, 7:40 PM

#

yep that's what people have to do currently with extdb for example.
But I don't want you to waste cpu time by checking every frame

#

Performance with intercept wouldn't be as bad as extdb if you'd actually do the checking. But still, 0 is better than something

cinder meteor Jan 11, 2019, 7:47 PM

#

so, am I right that with Intercept I can make such a thing:
make an optional addon which would replace some of my SQF functions with their intercept-based implementations, keeping the interface(parameters and their types) of the functions the same?

glossy inlet Jan 11, 2019, 7:48 PM

#

yeah. Intercept_CBA has a utility function where you can replace a SQF function (that can be call'ed) directly with your c++ function

cinder meteor Jan 11, 2019, 7:49 PM

#

damn it's nice... i was going to use some A* but it's quite slow with SQF. thanks.

glossy inlet Jan 11, 2019, 8:36 PM

#

Async was easier than I expected. The unscheduled one..
Okey.. What next... db_copyQuery done. db_waitForResult done
Okey... Have to rudder back, won't implement the proposed insert command. That's just inefficient, have to first call to the DB to check which columns are available then have to make a new query with that info.. Not worth the time investment right now.. I won't say no to a PR tho.
Rather invest the time into scheduled stuff.

smoky halo Jan 11, 2019, 8:39 PM

#

db_makeMeACupOfTea

cinder meteor Jan 11, 2019, 8:48 PM

#

An HTCPCP-compliant database?

glossy inlet Jan 11, 2019, 9:27 PM

#

Ah I love Generic Error in Expression without any info about what the error is :u

glossy inlet Jan 11, 2019, 10:02 PM

#

Wow..
Let's push the result onto the stack which is a ref counted smart pointer.
"Mh.. I don't know what this is.. But I can implicitly convert it to bool! Let's do that and push the bool to the stack!" 🤦

#

But atleast I now learned how to make async scheduled callstack thingies.
I could now make my script debugger work in scheduled 🤔

#

[] spawn {
    _query = db_prepareQuery 'SELECT intel_missionen.`name` FROM intel_missionen WHERE intel_missionen.id > ?';
    _query db_bindValue 400;

    _start = diag_tickTime;
    _result = DB_Connection db_query _query; //suspends script until result is ready
    systemChat ["got result!", diag_tickTime - _start];

    DB_RES = [db_resultToArray _result];
};

This makes it seem so much more professional. And it's so logical too

And it's a good step for Intercept. Before you could only have commands that freeze the game until they return, even if scheduled.
Now there is a example implementation for commands that just suspend in scheduled.

Build: https://github.com/intercept/intercept-database/releases/tag/0.3

scenic canopy Jan 11, 2019, 11:59 PM

#

nifty 😃

#

ported half my AAR thingy to intercept now finally

#

just need to add fire events etc too

sacred flume Jan 12, 2019, 1:13 AM

#

mmmm, I was going to write my own db connection, but if ur doing it already lewl

#

I was worried extdb3 would bog down too much if I increase polling, so this will be nice

#

600 games, 27GB of data so far :)

#

ill try swapping our mission over to it and see how it goes this weekend on test machine

sacred flume Jan 12, 2019, 7:47 AM

#

intercept-db should really load connection details from a config file

#

I like to have a param toggle between connections to swap between dev db and prod db

#

I also do like the seperate txt config for all the sql statements, loads them from it. don't like having to reload mission/etc just to edit

glossy inlet Jan 12, 2019, 9:21 AM

#

intercept-db should really load connection details from a config file Yeah that is one of my tasks for today. Don't really know how to properly do it.
I'd really like to load all that stuff from Arma config. But that would require you to pack a mod. or loadFile would be nice, but would require filePatching.
Probably gonna end up using a yaml config in @InterceptDB folder.

I like to have a param toggle between connections to swap between dev db and prod db Swap in the config file? or ingame in script?
It would be db_createConnection "configName"

Also for queries it would then be db_prepareQueryFromConfig "configName".
And I'll make it so that all existing/stored prepared config queries will be refreshed when you do a db_updateConfig. So that you don't need to re-prepare.

Question: rename db_stuff to dbStuff ? I liked the better seperation, but people might prefer proper camel.

meager falcon Jan 12, 2019, 9:48 AM

#

Can we also pass some extra args for the callback?

I.e. you insert a vehicle to the database and want to store the insertid on that vehicle in the callback

glossy inlet Jan 12, 2019, 10:13 AM

#

yeah.
bindCallback [code, arguments]
_this is [_result, _args]

nocturne basin Jan 12, 2019, 12:42 PM

#

No
Keep the db_ prefix
I consider it the same as with diag

glossy inlet Jan 12, 2019, 12:58 PM

#

What about cam/ctrl/config/lnb/curator/map/task/tv/waypoint commands tho. That's a valid point

nocturne basin Jan 12, 2019, 12:59 PM

#

there is a main difference: those are everyday commands

#

and diag commands are a very specific subset

#

just like the db commands are

dusky dune Jan 12, 2019, 1:27 PM

#

I say no space

#

Because of those other groups you mentioned

#

Underscore*

glossy inlet Jan 12, 2019, 2:13 PM

#

new stuff
dbCreateConnection "configname"
dbPrepareQueryConfig "configname"
dbPrepareQueryConfig ["configname", [binding1, binding2, ...]]
dbPrepareQuery ["INSERT ....", [binding1, binding2, ...]]
dbReloadConfig

And I made the move to camelCase because more voices for it than against.
if you reload config, all config queries update themselves automatically, no need to call dbPrepareQueryConfig again. But the db connection doesn't change itself if you change the ip in config 😉

#

Example with a config

accounts:
 maindb: #production db, don't break things here!
  ip: 127.0.0.1
  username: root
  password: lulz
  database: production
  port: 3306 #optional
 testdb: #testserver
  ip: 127.0.0.2
  username: root
  password: lulz
  database: production
  port: 3306 #optional

statements:
 insertStuff: INSERT INTO table (a,b,c) VALUES (?,?,?)
 deleteStuff: DELETE FROM table WHERE a=?

_connection = dbCreateConnection "testdb";
_query = dbPrepareQueryConfig ["deleteStuff", [15]];
_connection dbExecuteAsync _query;

I think dbExecute is better than dbQuery, more readable right?

nocturne basin Jan 12, 2019, 2:32 PM

#

there actually is a difference between execute and query for db

glossy inlet Jan 12, 2019, 2:33 PM

#

which is?

nocturne basin Jan 12, 2019, 2:34 PM

#

SQL has two parts

#

one is the query language (select etc)

#

second is the insert update etc. stuff

#

need to look up the exact names for those though ...

#

difference is: execute is usually done for insert, update and delete + all other operations that do not return a set of data

glossy inlet Jan 12, 2019, 2:35 PM

#

Execute query is more readable than query query though

nocturne basin Jan 12, 2019, 2:35 PM

#

query is used for sets of data

glossy inlet Jan 12, 2019, 2:35 PM

#

And so.. Execute is actually correct in most cases

#

I won't add a query command that does the same as execute just to have the correct names 😄

nocturne basin Jan 12, 2019, 2:35 PM

#

dbExecute and dbQuery would be the correct thing 🤷
one returning just the affected rows, the other the actual set

native kiln Jan 12, 2019, 2:39 PM

#

@nocturne basin DQL and DML ?

nocturne basin Jan 12, 2019, 2:40 PM

#

exactly

#

and Dedmens argument against query query has its roots in wrong naming in first place
dbPrepareQuery should be named dbPrepareCommand or dbPrepareStatement
then DBSET dbQuery ... and SCALAR dbExecute ... make more sense

glossy inlet Jan 12, 2019, 2:42 PM

#

yeah I name some things statement internally to. As in the config

glossy inlet Jan 12, 2019, 6:54 PM

#

Let's look at performance

dbCreateConnection "maindb"; 0.0013ms (connection is only established at first query)
dbPrepareQueryConfig ["getMissionName", [400]]; 0.0021ms
dbPrepareQueryConfig "getMissionName"; 0.0008ms
dbPrepareQuery "SELECT intel_missionen.name FROM intel_missionen WHERE intel_missionen.id > ?"; 0.0008ms (to be expected)
dbPrepareQuery ["SELECT intel_missionen.name FROM intel_missionen WHERE intel_missionen.id > ?", [400]]; 0.0021ms (to be expected, It's does the same stuff as the config variants)
DB_connection dbExecute DB_query 59.76ms well my ping to the server is 20ms..
DB_connection dbExecuteAsync DB_query; 0.0349ms

I like that speed for async.. I wonder how it compares to extdb.

fallen stone Jan 12, 2019, 8:04 PM

#

For the sync, not even close

#

For the async (using very poor SQF boilerplate to poll) - close

smoky halo Jan 12, 2019, 9:02 PM

#

@glossy inlet why not instal MySQL locally?

glossy inlet Jan 12, 2019, 9:15 PM

#

Because I don't need it ^^
I just need to test if it works. And I already have a big database with loads of test data to test on

#

Besides the actual statement execution the perf on sync and async should be about the same.
async still feels a little slow there is probably something I can optimize. Probably the worker thread doing the work interfering with the inserts to the queue as it has to lock the queue when taking out a task.

karmic niche Jan 12, 2019, 10:51 PM

#

won't implement the proposed insert command. That's just inefficient, have to first call to the DB to check which columns are available then have to make a new query with that info.. Not worth the time investment right now..
Uhh... I don't get it. Why did you have to check which columns are available? Why not just pass what the dev passed to the function?

#

Meaning: if you're passed positional arguments, just pass them as positional arguments without using column names.
If you're passed named arguments, just use those column names in your INSERT

smoky halo Jan 12, 2019, 11:03 PM

#

Are you using DB much?

karmic niche Jan 12, 2019, 11:58 PM

#

Yes. Care to explain? INSERT INTO table_name VALUES (value1, value2, value3, ...); does work so I don't see the reason for being passive-agressive.

glossy inlet Jan 13, 2019, 12:00 AM

#

does work Ah... Didn't know and didn't have time to google ^^

karmic niche Jan 13, 2019, 12:03 AM

#

Although if I wanted to play by your rules, I should have answered Yes. @smoky halo

smoky halo Jan 13, 2019, 12:05 AM

#

INSERT INTO table_name VALUES (value1, value2, value3, ...); cant find this syntax, got a link?

karmic niche Jan 13, 2019, 12:05 AM

#

https://www.w3schools.com/sql/sql_insert.asp

#

Although, as I have stated earlier, I'm almost always using ORM, tbh

#

And yes, of course, good practice would be to name the columns while inserting, hence the suggestion to use named arguments (["column", value]), which doesn't require you to do any check either

smoky halo Jan 13, 2019, 12:09 AM

#

Nope still can't find it https://dev.mysql.com/doc/refman/8.0/en/insert.html

#

But it works

karmic niche Jan 13, 2019, 12:16 AM

#

That's because square brackets in that link you posted are optional

#

Meaning INSERT square_brackets_are_optional VALUES (5); is valid MySQL syntax

#

(didn't even know that INTO was also optional; learned that today by looking at your link)

smoky halo Jan 13, 2019, 12:21 AM

#

Yes you can drop it, works without INTO

#

even on 5.7

#

VALUES or VALUE both work for multiple insert

fallen stone Jan 13, 2019, 1:45 AM

#

INSERT INTO table_name VALUES (value1, value2, value3, ...); cant find this syntax, got a link?

Seems you're not using DB much 🙃

sacred flume Jan 13, 2019, 2:59 AM

#

technically this is using the mariadb connector, so we should be using its documentation for reference :)

glossy inlet Jan 13, 2019, 5:08 PM

#

DB_connection dbExecuteAsync DB_query; 0.0036ms
That's more like it!

Btw don't run 10k async queries that actually do something when you profile.. Your database server will hate you
Build is up https://github.com/intercept/intercept-database/releases/tag/0.4

scenic canopy Jan 13, 2019, 5:25 PM

#

Connection pool? 😉

glossy inlet Jan 13, 2019, 5:27 PM

#

Just one worker thread for now. Per account, with it's own connection in the worker

scenic canopy Jan 13, 2019, 7:59 PM

#

thanks for the yaml example btw 😄

torn fulcrum Jan 14, 2019, 5:20 PM

#

Anyone ever created a battle eye whitelisting tool pulling data from a MySQL database? Thinking of using BattleNET

karmic niche Jan 14, 2019, 5:24 PM

#

BattlEye Whitelisting? You mean whitelisted?

torn fulcrum Jan 14, 2019, 5:28 PM

#

I mean using battle eye to kick players by checking their guid against a MySQL database. I didn’t word that in the best way 😔

low gyro Jan 18, 2019, 10:03 PM

#

//

graceful compass Jan 19, 2019, 4:12 PM

#

Does anyone here use Atom in combination with the sqf-linter from Lord_Golias? And does findIf throw an error for you or does it work as expected?
I get an error saying that findIf is a variable, when it should say it is a keyword. I have tried reinstalling it a couple times, but with no success.

pearl beacon Jan 19, 2019, 4:35 PM

#

update your sqflint package via pip

graceful compass Jan 19, 2019, 4:40 PM

#

I did that, using python -m pip install -e . in the my local git repo folder of the linter

pearl beacon Jan 19, 2019, 4:42 PM

#

make sure Atom's implementation is using latest version then

graceful compass Jan 19, 2019, 4:53 PM

#

got it fixed. Uninstalled everything, made sure all files related to it were deleted, redid the clone of the repo and installed.

pearl beacon Jan 19, 2019, 4:55 PM

#

👍

smoky halo Jan 20, 2019, 12:55 PM

#

hello

#

I'm new to arma modding

#

I want to make a tool that can connect to a server and get the scores or kills or something of players on it. Basically, some way of automatically communicating data out of arma

nocturne basin Jan 20, 2019, 1:04 PM

#

Cool @smoky halo
Noe, what is your question?

cobalt stag Jan 20, 2019, 1:29 PM

#

New RCon tool? It will be interesting

#

Just configured the Discord bot https://github.com/Windmolders/Z-Bot-Light/releases/tag/1.0 to interact with my server. I like it. Some features are not working as expected, but it almost that I need .

smoky halo Jan 20, 2019, 1:42 PM

#

@nocturne basin Is there anything already made that does that, or anything in that direction? I couldnt find anything by googling

#

.

nocturne basin Jan 20, 2019, 1:54 PM

#

Chances are, yes
You probably Look for some ist rcon Tool

smoky halo Jan 20, 2019, 2:11 PM

#

@smoky halo would be easier to make server send stats elsewhere at the end of round than connect to the server and query it

#

Also please do not crosspost

#

oki, gonna switch here

#

whats the easiest way to have server do that?

#

Make an extension https://community.bistudio.com/wiki/callExtension

#

alright

scenic canopy Jan 20, 2019, 5:11 PM

#

there's a couple of different tools that log game data

#

OCAP, R3 AAR etc

#

we have our own at Anrop, https://aar.anrop.se

#

the current main release of OCAP has a bunch of backdoors and security exploits so beware of that

orchid shadow Jan 20, 2019, 6:21 PM

#

@scenic canopy you got that on github?

scenic canopy Jan 20, 2019, 6:31 PM

#

yep, the current version, not the intercept one

#

or well, it's also on github but not public 😛

#

all of the mentioned above are on github

orchid shadow Jan 20, 2019, 6:40 PM

#

How much is the gain using intercept?

scenic canopy Jan 20, 2019, 7:03 PM

#

a lot easier to maintain

smoky halo Jan 20, 2019, 7:34 PM

#

oh thats nice

#

im good at python, so the more stuff i can do outside arma, the better for me

karmic niche Jan 20, 2019, 7:40 PM

#

@smoky halo If you're good at python, then here's some shameless self-promotion for you ;)
https://github.com/overfl0/Pythia

#

I really need to update the readme file, finally, but I'm too lazy to do that

smoky halo Jan 20, 2019, 7:53 PM

#

I need to make a bot for discord that awards points based on ingame actions

#

I'm scouting the grounds trying to figure stuff out

karmic niche Jan 20, 2019, 7:58 PM

#

You'd need to be able to connect discord players with steam ids, first.
Then: https://discordpy.readthedocs.io/en/rewrite/ 😃
(use the rewrite branch!!!)

smoky halo Jan 20, 2019, 7:58 PM

#

well, yep, usernames ingame

#

i know discord bots are py based

#

im trying to figure out if there is an easy way to do the arma info getting

#

like, what arma logs

#

etc

#

just found out about AAR

#

so, does it work with mods

#

like A>CE

scenic canopy Jan 20, 2019, 8:25 PM

#

which one?

#

and what does work with mods mean? 😛

karmic niche Jan 20, 2019, 8:27 PM

#

i know discord bots are py based
No they aren't :D
There are lots of bots written in discord.js, for example, but since you said you know python, I gave you the link to the python library and suggested the right branch, so that you don't waste your time

smoky halo Jan 20, 2019, 8:40 PM

#

like, there is the ACE medical system

#

can it log that stuff

scenic canopy Jan 20, 2019, 9:09 PM

#

sure, ACE emits events, such as unconscious or treatment

#

just like ACRE or TFAR emits events for radio broadcasts

smoky halo Jan 21, 2019, 8:01 AM

#

Oki, gonna take some time to research, a scoring system that works with ace would be a nice goal

scenic canopy Jan 21, 2019, 8:31 AM

#

@smoky halo this shows kills for ace in debriefing, https://github.com/acemod/ACEX/tree/master/addons/killtracker

GitHub

acemod/ACEX

ACE3 Extra misc modules and components. Contribute to acemod/ACEX development by creating an account on GitHub.

smoky halo Jan 21, 2019, 4:51 PM

#

thanks :)

smoky halo Jan 21, 2019, 9:18 PM

#

Which PBO tools do you guys use? I've used winse pbo manager 1.4 almost exclusively for testing but recently .net broke the right click > pack into pbo options for me. Is there any tool that offers the same ease of use that's also free? Alternatively if anyone has had the same issue and know a fix. I think my .net packages are more recent and somehow not compatible :/

scenic canopy Jan 21, 2019, 9:40 PM

#

armake(2) and mikero tools

native kiln Jan 21, 2019, 9:47 PM

#

@smoky halo you can simply readd it to your context menu

smoky halo Jan 21, 2019, 11:18 PM

#

@native kiln how? Wouldn't the program not know what to do?

native kiln Jan 21, 2019, 11:18 PM

#

does your pbo manager still work by itself? as in, you can run the program via the exe?

smoky halo Jan 21, 2019, 11:19 PM

#

Most of the time it works without errors 😉

native kiln Jan 21, 2019, 11:21 PM

#

so i assume youre just missing the context menu click on folders (make to pbo) and vice versa?

#

in that case, have you tried just reinstalling pbo manager?

#

that should re-register the context menu entries

smoky halo Jan 21, 2019, 11:24 PM

#

Yep, doesn't work. Since it's an EXE I can't really do much about it. Wouldn't know how to make a context menu addition to work with it neither.

native kiln Jan 21, 2019, 11:32 PM

#

Hm, you can fix it with the registry editor, however I don't know if you feel comfortable doing that

smoky halo Jan 21, 2019, 11:33 PM

#

I know enough to not break it. Not sure how you'd fix that though. Feel free to hold my hand if you want 😃

nocturne basin Jan 21, 2019, 11:36 PM

#

https://www.thewindowsclub.com/remove-click-context-menu-items-editors
literally first entry when looking for it https://www.google.com/search?q=add+context+menu+entry+windows+10&rlz=1C1CHBF_deDE718DE718&oq=add+context+menu+entry&aqs=chrome.1.69i57j0l4.3520j0j4&sourceid=chrome&ie=UTF-8

native kiln Jan 21, 2019, 11:36 PM

#

First website's tools don't work properly

nocturne basin Jan 21, 2019, 11:36 PM

#

no need for the tools though

native kiln Jan 21, 2019, 11:37 PM

#

I've got my experience with these tutorials telling you what keys to create

#

most of the time they just didn't work, don't ask me, ask Windows

#

¯_(ツ)_/¯

#

and sadly the helper tools didn't detect all programs correctly

smoky halo Jan 21, 2019, 11:37 PM

#

IDK what to do about pbo manager in there though

smoky halo Jan 21, 2019, 11:54 PM

#

For future reference pbo manager has a seperate 32 bit installer which properly installs a key. Not sure why 64 bit didn't function for me.

native kiln Jan 21, 2019, 11:55 PM

#

Remembered that was the same issue for me some months back 😄

dawn palm Jan 22, 2019, 4:15 AM

#

pboManager is an excellent, intuitive, and easy-to-use tool >for missions<. It can't realistically be used for (most) addons because there's no facility in it for the pbo prefix. It's able to read the prefix, but cannot create one.

#

extractpbo (gui) is similar to pboManager in functionality, but a little more complex due to the fact that it does cater for prefix.

#

the other problem with the tool you like is it cannot binarise/unbinarise configs,. missions, or rvmats so is somewhat useless (especially for extraction)

#

I don't think Armake has a gui that's suitable for you, but if it does, use it. Unlike bis, if you discver a bug in Armake, it's likely to get fixed.

tight cypress Jan 22, 2019, 5:17 AM

#

You can create a prefix with pboManager, it's just not intuitive at all.

sick verge Jan 22, 2019, 5:23 AM

#

Does someone know whether the game performs more than just packing a mission as pbo when exporting it in the editor?
Because I tried to manually pack the mission folder as a pbo via armake (no additional includes or anything) but the resulting mission-pbo just doesn't work properly...

dawn palm Jan 22, 2019, 5:52 AM

#

the exported mission.sqm is plain jane text, nothing wrong with that, but binarise is better (and I imagine armake would do that automatically)

#

it's more likely to be something you've done wrong/missing with the description.ext

sick verge Jan 22, 2019, 6:28 AM

#

During export? I left armake with the default settings - not quite sure whether that does anything with the description.ext

#

So I only used armake build <MissionFolder> <pboName>

#

And without chasing anything in the mission it works when exporting from the editor 🤷

vague shard Jan 22, 2019, 6:37 AM

#

@sick verge did you compare the resulting pbos (extracted to source again without derap)?

sick verge Jan 22, 2019, 6:41 AM

#

not yet - but I gues I should do that. Just thought this was kind of a known problem or something like this

vague shard Jan 22, 2019, 6:45 AM

#

compression for sqm is broken in A3

#

been a while i last checked

sick verge Jan 22, 2019, 6:52 AM

#

Well the mission.sqm I am using is binarisied already so that shouldn't make any difference

dawn palm Jan 22, 2019, 7:08 AM

#

well apart from the sqf code, what's so special about a mission? even addon breaker can make them. so why would armake or even my tools be needed? perhaps you should try pboManager first which lets you 'make' a pbo that is a mirror image of your current folder.

sick verge Jan 22, 2019, 7:11 AM

#

Will try

glossy inlet Jan 22, 2019, 7:41 AM

#

@sick verge armake might try to binarize the description.ext by default which is kinda wrong

#

Or could you explain what you mean by "doesn't work properly" ?

pearl beacon Jan 22, 2019, 9:35 AM

#

armake works fine for packing missions, used it quite a bit

sick verge Jan 22, 2019, 9:41 AM

#

Interesting - "Doesn't work properly" means that either

When starting the mission (built on Stratis and saved with the respective extension) I suddenly stand in the middle of Tanoa's jungle (all items are correctly placed around me though - so it appears to be just the map)
Or: I start the mission and it immediately ends the mission as failed

Seems very odd me. Both times the issue went away when exporting it from ingame 🤷

#

Maybe I should add that I am using a Linux-server - this could make a difference too

glossy inlet Jan 22, 2019, 9:56 AM

#

is the filename correct then? mission.stratis.pbo?

sick verge Jan 22, 2019, 11:02 AM

#

yeah that's what I meant with "proper extension"

#

anyways - I will compare both versions and see if I can find any difference when unpacking them again

glossy inlet Jan 22, 2019, 11:02 AM

#

and saved with the respective extension Ah I overread that

pearl beacon Jan 22, 2019, 11:02 AM

#

Interesting, I was using my own build, so it could become broken on later ones

glossy inlet Jan 22, 2019, 11:03 AM

#

If you can't find the difference just send me a working and non working pbo. I'll find it

sick verge Jan 22, 2019, 11:03 AM

#

Alright thanks 😃

glossy inlet Jan 22, 2019, 5:29 PM

#

https://s.sqf.ovh/chrome_2019-01-22_18-29-28.png New frontend for my profiler incoming.
I also now know how I could display scheduled scripts too. Even though their data wouldn't be that useful.. cuz you know.. scheduled

glossy inlet Jan 22, 2019, 6:02 PM

#

Next up is implementing support for "Tracy" profiler.
Which means live-capture over network, and memory efficient enough to probably be able to record a whole hour of session.
Also if you have a dokanPbo or normal P drive you should be able to open source-files belonging to the "zones" directly.
https://www.youtube.com/watch?v=fB5B46lbapc Here is a introduction video to Tracy.
This will be the next evolution of the profiler.
Maybe I can even add the Arma profiling build engine profiling to that. That would make it REALLY powerful.

Tracy also let's you directly compare old and new captures so you can easily see if your changes actually improved performance https://s.sqf.ovh/firefox_2019-01-22_19-07-50.png
This is literally making me 🤤

orchid shadow Jan 22, 2019, 9:06 PM

#

Dammit dedmen, you crazy! but i love it!

buoyant citrus Jan 22, 2019, 9:12 PM

#

this is gonna expose my shitty programming

#

gonna stop me blaming BI

glossy inlet Jan 22, 2019, 9:15 PM

#

I could add things like current number of vehicles or AI groups and such to that..
Make it be like arma server monitor just alot more overkill :D
Object count would be useful for diagnosing the terrible FPS drop I had in last mission when the cold breath script left like 20k particle emitters floating around..
Though if I add engine profiling that would also show up in there.. I hope

buoyant citrus Jan 22, 2019, 9:16 PM

#

could you add a headless client watcher to see the effects headless client has on the server

#

anything to help debug is helpful

glossy inlet Jan 22, 2019, 9:18 PM

#

what would it watch?
It can already profile scripts on dedicated clients.
But how would it see if HC has positive effect?
Besides the server fps being higher.. But that's way easier to see by just using the #monitor chat command

#

It will tell you where your script bottlenecks are tho. And you might be able to figure out yourself that you can move that stuff from server to HC to get better perf

buoyant citrus Jan 22, 2019, 9:21 PM

#

yes but a server monitoring can show the fps being 50 but the AI is de-syncing all over the place,
It be nice to see the HC is having a positive affect, especially when units are being spawned in plus if blacklisting units from HC is beneficial or not

glossy inlet Jan 22, 2019, 9:23 PM

#

Don't really know how.. But then.. Most of this stuff is about things that people never thought of doing previously sooo... 😄

buoyant citrus Jan 22, 2019, 9:27 PM

#

hey bud thats what its all about. It doesn't matter if you cant i was just spit-balling random crap.
Plus you're doing gods works. Finally being able to see what is killing performance instead of blaming BI plus finally be able stop all the arguments about what is the best method for doing certain things.

glossy inlet Jan 22, 2019, 9:28 PM

#

Hah 😄 The second part is why I originally started with this 😄

#

use private keyword bois!

pearl beacon Jan 22, 2019, 9:29 PM

#

but the second part was doable before, it just had to be much more granular

buoyant citrus Jan 22, 2019, 9:29 PM

#

i like to go with my method is best the bad performance is BI, dont @ me about doing everything in PFM.
sidenote does this have any benefit to modders with LOD and models.

#

yes but who has time for that when you can blame BI?

glossy inlet Jan 22, 2019, 9:29 PM

#

The first part too with profiling build.
Just alot more granular if the cause of problems were scripts

#

no this does nothing for modeling and stuff.

buoyant citrus Jan 22, 2019, 9:30 PM

#

also your script could be being blocked by something else

glossy inlet Jan 22, 2019, 9:30 PM

#

This is 99% scripting

buoyant citrus Jan 22, 2019, 9:30 PM

#

i thought so, just checking

glossy inlet Jan 22, 2019, 9:33 PM

#

If you are interested in modelling stuff you could get a real GPU profiler like NView thing from Nvidia.. Not sure what the exact name is, but I have it... Nsight! right! Nvidia Nsight.
I could probably add more fine grained engine profiling to for example log how long each model takes to draw... But... I have no need for that and few people do. So I probably won't do that.

buoyant citrus Jan 22, 2019, 9:35 PM

#

me, modelling lmao. im a programmer if its anything more than a block with a flat single texture for testing it aint happening, I was just wondering for performance testing of mods

#

i'll leave that to the artist of this community

glossy inlet Jan 22, 2019, 9:36 PM

#

Well if there are script problems in mods it will light up bright red :D
Found many ACE/TFAR problems (and one huge bwmod one) with that already ^^

buoyant citrus Jan 22, 2019, 9:36 PM

#

soo much easier but not as flashy in end compare to a tank or jet

#

but i was more on about textures and models

#

for example some weapons packs love to eat fps

pearl beacon Jan 22, 2019, 9:47 PM

#

Can't wait for semester to finally be over lol

#

Find some stupid performance eaters

#

But I am saying that now because I don't want to study, and after I won't have to for a bit, I'll even be too lazy to profile stuff

smoky halo Jan 22, 2019, 9:48 PM

#

I used Nvidia Nsight before to profile Arma and it confirmed a few useful things generally like the drawcall count and overhead of doing so is relatively low but it didn't give me a great deal of information beyond that without the source code for the game.

#

I can see it might be useful for something like model testing though if you can compare scenes with and without the models you have added, but I don't recall getting the sort of granularity that you could use for directly attributed the time your thing does (but I wasn't admittedly using it for that, I was trying to ascertain if DX12 would have any chance of improving Arma performance after they announced they were trying it)

dusky dune Jan 22, 2019, 9:50 PM

#

Finally being able to see what is killing performance instead of blaming BI
This is the 2nd level meme though

#

There's still tons of things where you can 100% blame BIS for performance

#

Particles being one of them

#

Smaller cell size terrains

smoky halo Jan 22, 2019, 9:52 PM

#

I think Particles is now the big problem we see, but I also think that is also being impacted by ACEs cookoffs and if you add blastcore and such that gets way worse. The profile build doesn't really show it up very well either. I think there is a leak of particles that grows over time but I haven't yet gathered sufficient evidence to point directly at it

#

Also when can I get my hands on this amazing profiling tool? You know I want it!

buoyant citrus Jan 22, 2019, 9:53 PM

#

i think everyone wants its, even bi

#

but they probably have their own inhouse one

#

and as for it being a 2nd level meme i know.But at least for me a lot of stuff i see is normally cause by people doing really dumb shit and blaming BI. But other times its BI fault and its just, work around it or deal with it

smoky halo Jan 22, 2019, 9:56 PM

#

If you want a beta tester I am happy to help as always. The last version I still use to test my scripts, just a shame it never directly tracked down that antistasi issue but I think whatever that is seems to be entirely with BI and we don't quite enough information on what could cause it

pearl beacon Jan 22, 2019, 9:57 PM

#

Terrain cell size and other quality stuff like that will always be a tradeoff between performance and quality though

smoky halo Jan 22, 2019, 9:58 PM

#

Certainly need to have a look into the CF_BAI/detect routines, 30 v 30 is taking way too long currently despite all I have done to reduce calculation and cache results.

#

We reject quite a few terrains that just run too poorly, they look lovely in images but when you don't get good frame rates on an empty terrain there is no way people are going to enjoy it when its pull of enemies and the fps is even lower. I can see why a maker makes that trade off but I think performance is a really important part of what people look for in terrains at a minimum level of quality.

buoyant citrus Jan 22, 2019, 10:00 PM

#

yeah i agree, i'd rather play on a a1 map with 60+ fps then a lot of maps with 20

#

its hard to enjoy the scenery when im looking at it in a powerpoint

dusky dune Jan 22, 2019, 10:01 PM

#

Terrain cell size and other quality stuff like that will always be a tradeoff between performance and quality though
Sure, but BIS one isn't really on the high end there 😄

#

Enfusion rumors are better there

#

Like you get people blaming BIS for everything, but also thinking they're holy and vanilla FPS is amazing

#

which really it's not either, it's very easy to get bad performance in arma. Heck look at Argo, performance is trash in it and it's fully within BIS own control

buoyant citrus Jan 22, 2019, 10:02 PM

#

nah, have you been in kavala on a summers day

smoky halo Jan 22, 2019, 10:03 PM

#

Kavala performance has improved quite a bit from where it was a few years ago. Still the worst performing point on Altis but above 60 there is now consistent everywhere.

buoyant citrus Jan 22, 2019, 10:04 PM

#

but i also agree a lot of hate towards bi is down to bad mods and mission makers being like 100+ ai.
i would like to see improved core utilisation but from what i know and read thats more an engine issue

pearl beacon Jan 22, 2019, 10:04 PM

#

I am not saying BI isn't to blame, just stating the obvious
Also BIS is not BI

buoyant citrus Jan 22, 2019, 10:05 PM

#

you still know what i mean, it doesnt matter for an extra letter

dusky dune Jan 22, 2019, 10:05 PM

#

im so used to saying bohemia interact studios

pearl beacon Jan 22, 2019, 10:05 PM

#

You can even run 200 AI if you know what you are doing and get good performance
It's just that most people simply don't know how any of it works, and there are a lot of people spreading lies

#

So more people read those and follow without thinking

buoyant citrus Jan 22, 2019, 10:07 PM

#

i know about that but still having 100+ ai all in active combat near a player is a no no and some people blame bi for that.

#

putting some dynamic simulation with some manual simulation with syncing groups can allow for much greater numbers on preloaded AI along with HC

#

but how many of the arma playing base know about all the tricks to improve mission performance.
im not sure but i think this is getting off topic

dusky dune Jan 22, 2019, 10:11 PM

#

tricks to improve mission performance.
The tricks BI doesnt evne use in their own missions

pearl beacon Jan 22, 2019, 10:11 PM

#

Not many, and that's a problem

dusky dune Jan 22, 2019, 10:11 PM

#

Like I know there's a ton of tricks, but a lot of pure vanilla stuff isn't great

pearl beacon Jan 22, 2019, 10:11 PM

#

And yes, offtopic :p

dusky dune Jan 22, 2019, 10:11 PM

#

Smoke can be optimized quite a bit (Both in actual usefullness, lol at vanilla rubbing values)

#

But.... it's the usual arma thing of relying on the community to fix problems

buoyant citrus Jan 22, 2019, 10:12 PM

#

well lets wait and see what A4 has to offer

pearl beacon Jan 22, 2019, 10:13 PM

#

Good joke

dusky dune Jan 22, 2019, 10:14 PM

#

Well at least for now dayz performance is quite a lot better

tight cypress Jan 22, 2019, 10:34 PM

#

More off topic comments but when people remember OFP and even ArmA 1 we could have a ton of AI and performance was not as degraded as it was in A2/A3 but of course with much lower quality AI processing. We could have tons of mods in OFP and it impacted performance to a much lesser degree probably because of the simplicity of all the assets and the general effectiveness of the engine. (at least on a fast computer) That was one of the big innovations of both Source and RV... they were efficient enough to allow novice scripters to use that performance for their mods, and it is no surprise they have large modding communities. Hopefully Enfusion can be efficient enough to give us plenty of frames for mods to use for useful things, it should be a high priority. 😃

pearl beacon Jan 22, 2019, 10:52 PM

#

you have a point there, that's what happens when you use the same engine but don't or can't update it enough to use newer hardware as efficiently as it did when it was new

smoky halo Jan 22, 2019, 11:37 PM

#

It is one of the reason I want to get access to more cores with the next engine. I want to be able to explicitly utilise threads. Since the entire world of CPUs is clearly heading towards more cores, and it is vastly cheaper to add cores than it is more clockspeed or IPC this is what a future architecture requires. The core game might not utilise those ores but as a modder I can often be very parallel and we could do a heck of a lot more if we had that ability with Arma 3. As is we are so constrained by how little performance is left that it diminishes what is possible while having most of everyones CPU sitting idle doing nothing

glossy inlet Jan 23, 2019, 7:46 AM

#

@smoky halo Also when can I get my hands on this amazing profiling tool?
https://github.com/dedmen/ArmaScriptProfiler/ Current version supports Brofiler or Chrome. The new Tracy stuff is only coming today.
but they probably have their own inhouse one Jup. It's called diag_captureFrame And we happen to have access to that one too.
Btw didn't notice any particle problems. Besides that script that spawned thousands of emitters last mission, but the problem cause was the thousands of emitter vehicles, not the particles.

glossy inlet Jan 23, 2019, 9:36 AM

#

Woah nice.. Tracey also captures crashes, combined with engine profiling you could see in which region the engine was when it crashed. If it was in model loading or rendering you know something is probably wrong with a model.
That will be nice if that really works. Not ultra useful, but a nice extra.

Can also display diag_log calls in the trace, so you can see which function called diag_log at what time..
Don't think I wanna do that for engine RPT logging too, too much work. But I could.

How about a memory allocation/useage graph? that is fairly "trivial" to do. Don't know if that would be really that useful. But if you see that your script is allocating tons of memory that would be very useful. Could help to detect memory "leaks" caused by scripts.

Okey it seems I have the basic Tracy implementation done, was easier than expected.

Tracy also has a command line application that just records data to a file on disc if you don't wanna view it live. Man I need to write a wiki page for this.

glossy inlet Jan 23, 2019, 11:16 AM

#

Holy macaroni.. My hype levels have reached the 9000 mark.

📎 2019-01-23_12-14-56.mp4

#

All that is running from my Home PC to my work PC over a Teamviewer VPN.

tight cypress Jan 23, 2019, 11:22 AM

#

statemachine checks are exactly what I am excited to use with that tool when its ready, great job

glossy inlet Jan 23, 2019, 11:22 AM

#

After 11 minutes the ram usage is at 85MB.
Want to see what functions take the most time, over a long period session?
Sure! here you go!

📎 Tracy_2019-01-23_12-22-17.png

#

This is literally the best thing ever

#

Currently it doesn't log _this arguments. I had that in the Brofiler version, But that really has no place in a profiler so I won't re-add that for the Tracy frontend.

meager falcon Jan 23, 2019, 11:27 AM

#

Clockwork really hits it off in this one 😅

glossy inlet Jan 23, 2019, 11:28 AM

#

10 seconds over 11 minutes is like 1.5%

#

Can't stop looking at this

📎 2019-01-23_12-31-30.mp4

tight cypress Jan 23, 2019, 11:35 AM

#

Senfo - 2019-01-23 03:27 -08:00

Clockwork really hits it off in this one :sweat_smile:

#

I accept the challenge of making a clockwork iteration be the most intensive. I'm converting this to statemachine
https://i.imgur.com/BZWvG7z.png

glossy inlet Jan 23, 2019, 11:38 AM

#

I wonder what happens when I enable instruction level profiling 🤔
I had to disable that for brofiler because it just lagged it to death.

pearl beacon Jan 23, 2019, 11:39 AM

#

CBA statemachine was (IIRC) designed so it will always have pretty much constant workload no matter how large the statemachine is

#

the statemachine handling itself of course, not the subfunctions

glossy inlet Jan 23, 2019, 12:04 PM

#

Instruction level profiling works splendidly.
compileFinal is calling compile because the profiler redirects it. And the gap between end of compile and end of compileFinal is the profiler inserting instrumentation instructions into the compiled code.

📎 Tracy_2019-01-23_13-01-58.png

smoky halo Jan 23, 2019, 12:05 PM

#

I have no doubt that the release of this new version is really going to help the community develop better scripts because there is a darn good chance we will find what is going slowly over the longer term. Looking incredible!

glossy inlet Jan 23, 2019, 12:06 PM

#

That stuff was already possible with the previous versions. Just not that feature rich and easy to use

smoky halo Jan 23, 2019, 12:07 PM

#

Better scripts? Doubt it very much, people who write shit scripts will continue to do so. The only thing this would help is to people like Dedmen to identify where BI fucked up

glossy inlet Jan 23, 2019, 12:12 PM

#

Want to know how much impact retriving variables has? Easy! Just click on the getVar instruction!

Man I'm so hyped..

📎 Tracy_2019-01-23_13-11-46.png

#

So far getting a local variable seems to be about 300-500ns
but a global variable can be up to 5us.

smoky halo Jan 23, 2019, 12:15 PM

#

Try object namespace for comparison

glossy inlet Jan 23, 2019, 12:15 PM

#

Not the getVariable command.
the get variable instruction.
Commands are all >1us

smoky halo Jan 23, 2019, 12:19 PM

#

Does setAmmo immediately broadcast ammo change of a unit or there is a scheduled update packet?

glossy inlet Jan 23, 2019, 12:19 PM

#

All packets are pushed into the queue immediately. The queue will be processed "later" like.. a couple milliseconds later maybe

smoky halo Jan 23, 2019, 12:20 PM

#

So if you setAmmo each frame you will have a train of packets?

#

Position sync has a scheduled update

glossy inlet Jan 23, 2019, 12:22 PM

#

Arma might batch them together as it sees fit. But if you didn't get close to the bandwidth limit it will just send them out

cinder meteor Jan 23, 2019, 1:14 PM

#

I was playing with concurrency analyzer for visual studio, and it has nice API for adding flags/spans/messages/whatever with custom text to the graph. I guess you can easily add similar SQF commands to your profiler which would interface with this Tracy program? 🤔

glossy inlet Jan 23, 2019, 1:58 PM

#

Already have SQF commands for the profiler.
You can dump messages and set counters so far.
Don't know what else I could add. measurement zones, messages and counters is about everything

cinder meteor Jan 23, 2019, 2:01 PM

#

Then it's great 😃
I thought it only plots stuff. Never tried yet because so much of my stuff is scheduled

#

I was looking through chat... somewhere you said that you can add profiling of scheduled scripts... can't find the exact message
I'd really love to see that. Is it mega-hard to do? I guess you'd just plot every spawned script as thread 🤷 Maybe add indication when arma scheduler suspends my scheduled script because it took all the time interval and suspends it, I don't know

glossy inlet Jan 23, 2019, 2:10 PM

#

It plots all script functions. And you can add your own mid-function measurements if you need.

Plotting every spawned script as thread is the easy solution yeah.
Finding out when it's actually being executed is a little harder. Still quite easy for my standards ^^ Just alot of work

smoky halo Jan 23, 2019, 2:11 PM

#

Why, what would be the use of it?

glossy inlet Jan 23, 2019, 2:11 PM

#

I work on the stuff that I find useful. And I don't care about scheduled so it's not high on my list

#

And I don't think the frontend would properly display that stuff either.

#

It won't be useful for performance analysis I'd say. But it would be useful to see what's going on right in that moment.

cinder meteor Jan 23, 2019, 2:12 PM

#

I'd say if I develop my stuff with scheduled scripts, I already abstract away the fact that arma stops it and suspends it when it think it's ok. I don't really care about that. What I care is how much it takes my thread to perform a task, irrelevant of FPS.

smoky halo Jan 23, 2019, 2:13 PM

#

Don’t understand

cinder meteor Jan 23, 2019, 2:14 PM

#

Hmm... I spawn a script as a sort of worker thread abstraction. I push messages to its queue to make it do work. I'd like to see how long it takes to process messages and plot that stuff. If arma suspends it in the middle of processing a message and resumes a few frames after that, I don't really care.

#

So, let's say I can start a span (this rectangle in the profiler) when I start executing my code in scheduled and stop it when I'm done. If the profiler figures out that the start_span was called from one script (thisscript), it should place this span into the proper horizontal line which corresponds to this thisscript, that's what I mean 🤷

smoky halo Jan 23, 2019, 2:19 PM

#

I’m afraid your script won’t resume in a few frames it could take much more than that in scheduled

cinder meteor Jan 23, 2019, 2:20 PM

#

But we don't care about that if it's scheduled, it's normal, all scheduled scripts do that 😄

smoky halo Jan 23, 2019, 2:21 PM

#

Sorry you are talking about profiler presentation?

cinder meteor Jan 23, 2019, 2:21 PM

#

yes

#

I'm talking about how profiler could in theory show it

smoky halo Jan 23, 2019, 2:22 PM

#

Ok not what I asked though

cinder meteor Jan 23, 2019, 2:24 PM

#

?

smoky halo Jan 23, 2019, 2:27 PM

#

Was curious how useful this could be for writing scheduled scripts. We already know why suspension happens and when, how seeing it drawn on graph would help?

cinder meteor Jan 23, 2019, 2:30 PM

#

Hmm... IMO profiler is good because it lets you show how the state of your program varies in time. And suspension is an important program state. So I think it would make sense to plot it. 🤔

#

@glossy inlet Where can I see what SQF commands your ArmaScriptProfiler provides?

smoky halo Jan 23, 2019, 2:32 PM

#

The 3ms is hardcoded if it was variable depending on FPS I would agree with you but it is a constant

cinder meteor Jan 23, 2019, 2:34 PM

#

Yes but you would see that your spawned script is definitely suspended now, and not just stuck/doing nothing. I guess having some sort of indicator when arma switches scheduled scripts should be useful for that. But even if there is no such indicator, being able to profile scheduled would be great.

glossy inlet Jan 23, 2019, 2:37 PM

#

https://community.bistudio.com/wiki/ArmaScriptProfiler
There is a profilerSetCounter command that's not documented "name" profilerSetCounter _number
And all the capture/trigger commands are for the diag_captureFrame frontend, which is garbage

#

The profiler would show the difference between a single scheduled script running the full 3ms and then being force suspended. Or it exiting early because of waitUntil or sleep or similar.
Could help identify while true garbage without sleep

cinder meteor Jan 23, 2019, 2:40 PM

#

Maybe you could add commands to create a line with specific name in the profiler plot, and to start/stop a span (this horizontal rectangle in the line... how is it called >_<) in a specific line? Then I'd integrate the profiler into my SQF code easily 😄

glossy inlet Jan 23, 2019, 2:40 PM

#

already exist...

#

create a line with a message profilerLog.
start/stop createProfileScope

cinder meteor Jan 23, 2019, 2:43 PM

#

So, I create a scope with createProfileScope, then the script suspends before the returned ProfileScope variable is destroyed. How will the profiler draw it?

glossy inlet Jan 23, 2019, 2:44 PM

#

not at all

#

stuff in scheduled is ignored

smoky halo Jan 23, 2019, 2:44 PM

#

How about sleep [time, count] command that returns bool especially for while loop garbage? So time is sleep time, count is how many times this command will return true before it returns false. -1 always returns true

glossy inlet Jan 23, 2019, 2:45 PM

#

wat

#

why not just use a for loop if you already know how many iterations you want. And then just place a sleep inside it?

smoky halo Jan 23, 2019, 2:45 PM

#

Just throwing some garbage ideas around

cinder meteor Jan 23, 2019, 2:45 PM

#

That's what I mean... If I could have an SQF wrapper API for drawing stuff in the profiler visualizer program, I'd easily profile my scheduled code, because the lines/spans/scopes would be made by me

smoky halo Jan 23, 2019, 2:46 PM

#

You can do that of course but where is fun in that?

cinder meteor Jan 23, 2019, 2:48 PM

#

I would see what's happening in my code easier... currently I have a few threads running and they dump stuff into the rpt, which is getting hard to read

smoky halo Jan 23, 2019, 2:48 PM

#

if !(sleep [1, 10]) exitWith {};

pearl beacon Jan 23, 2019, 2:49 PM

#

"threads"

cinder meteor Jan 23, 2019, 2:49 PM

#

If it behaves like a thread then I can call it a thread >_<

pearl beacon Jan 23, 2019, 2:49 PM

#

but it doesn't -.-

glossy inlet Jan 23, 2019, 2:50 PM

#

it behaves like a thread on a single core cpu

cinder meteor Jan 23, 2019, 2:50 PM

#

I could argue that there are no threads on a single core CPU, come on

pearl beacon Jan 23, 2019, 2:50 PM

#

well that's true lol

#

but with 3ms, it's hardly even that 😂

vague shard Jan 23, 2019, 3:28 PM

#

if this can log stats, then you could gather useful info on scheduled code

glossy inlet Jan 23, 2019, 3:35 PM

#

stats?

#

Man.. I had literally one byte too little space to plug in for the engine profiling..
So I had to write new hooking assembly that is now 3 2 bytes smaller than what I was previously using.
Revolutionzing my other projects while doing this :u

Also I would've had to leave 3 minutes ago to catch my bus... Guess I'll stay a little longer then 😄

cinder meteor Jan 23, 2019, 4:18 PM

#

So do my profiler visualizer wrapper functions make any sense or nah? 🙄

glossy inlet Jan 23, 2019, 4:19 PM

#

Maybe... But they won't work in scheduled.
And if I make them work in scheduled then.. We already have all that stuff..

cinder meteor Jan 23, 2019, 4:21 PM

#

Why won't they work? An sqf command asks Tracy to register a line where it will draw rectangles, and it just does it? Then SQF asks Tracy to draw this rectangle, etc. Or is it not how Tracy and your other profiling visualizers work?

glossy inlet Jan 23, 2019, 4:24 PM

#

Ah now I understood

#

yeah I can enable profilerLog command in scheduled. I think it already is enabled

#

that's not really how tracy works no.
I don't tell tracy to draw. I tell tracy when a zone starts and when a zone ends

#

I think I got engine profiling working. But I gotta catch my bus now.. Cannot test. Foof

cinder meteor Jan 23, 2019, 4:25 PM

#

Yes, what if I could tell tracy when a zone start and ends through SQF, that's what I am trying to explain 😄

glossy inlet Jan 23, 2019, 4:26 PM

#

Even if. Scheduled.. You have 0 control

#

I already have a solution.. Will probably be easier

#

Welp arma got to main menu without crashing.. Engine profiling probably works.. I'll present actual results when I'm home in 30

vague shard Jan 23, 2019, 4:59 PM

#

stats
total number of executions
executions per timeframe or specific time box
"parallel" executions

or stuff like execution tree

glossy inlet Jan 23, 2019, 5:04 PM

#

what does execution mean?

vague shard Jan 23, 2019, 5:08 PM

#

call/spawn (etc?) of a function

#

in simple terms of how often the code is ran (total/per timeframe/per timebox)

#

similar would be useful for network calls (pV, setVar public, remoteExec[Call])

glossy inlet Jan 23, 2019, 5:09 PM

#

could already enable instruction level profiling.
Atleast for remoteExec the command executions will show up. Also in scheduled scripts

#

https://vid.pr0gramm.com/2019/01/18/a3923b4e83e9f353.mp4
I got this running in a loop. Really helps me progress faster

▶ Play video

dusky dune Jan 23, 2019, 5:17 PM

#

doesnt loop perfect

#

0/10

glossy inlet Jan 23, 2019, 5:18 PM

#

someone make me a perfect loop so that I can work even faster 😄

#

But I can already see the finish line anyway. So maybe not worth it.
Gonna go full speed now! nobody gonna slow me down! o no arma crashed! I got to keep on movin!

glossy inlet Jan 23, 2019, 5:48 PM

#

https://s.sqf.ovh/Tracy_2019-01-23_18-48-42.png sooo close.. Zones don't exit when they should.

sick verge Jan 23, 2019, 6:47 PM

#

@glossy inlet You remember me complaining about missions being packed with armake not working? You were right with your suspicion that armake binarises the description.ext. So I guess that is the issue

glossy inlet Jan 23, 2019, 6:49 PM

#

I remember commenting on it that it binarizes everything that ends with ext which is plain wrong, if at all then only description.
But I don't remember ever seeing a binarized description.ext in a mission pbo

sick verge Jan 23, 2019, 6:50 PM

#

Pack it with armake v0.6.3 and you'll see one 😛

glossy inlet Jan 23, 2019, 6:54 PM

#

Fixed in armake++ ^^ (Don't build p3d's with this)

📎 armake.exe

sick verge Jan 23, 2019, 7:12 PM

#

Alright thanks 👍

#

What is armake++ though? A c++ port of the original? 🤔

#

Alright apparently the binarised description.ext was not the cause of the mission being faulty - the search continues...

glossy inlet Jan 23, 2019, 7:16 PM

#

yep

#

As I said. Send me working and non working pbo and I'll find the difference ^^

#

preferably tomorrow. Wanna finish the profiler stuff first

sick verge Jan 23, 2019, 7:17 PM

#

Will do as soon as I actually validated that the "working one" is actually working ^^

glossy inlet Jan 23, 2019, 8:28 PM

#

It's done!!!!

📎 2019-01-23_21-28-12.mp4

#

The sad thing about this is... I'm VERY sure I now have better profiling tools than even what BI uses internally...

obsidian sluice Jan 23, 2019, 8:30 PM

#

No

#

😛

glossy inlet Jan 23, 2019, 8:30 PM

#

Better stuff than diag_captureFrame?

obsidian sluice Jan 23, 2019, 8:30 PM

#

Yes?

glossy inlet Jan 23, 2019, 8:31 PM

#

If only the modders could have access to that 👀

obsidian sluice Jan 23, 2019, 8:31 PM

#

Not my call

#

shrug

glossy inlet Jan 23, 2019, 8:32 PM

#

Well now we atleast have a alternative..
A VERY ram hungry alternative :D
3GB in 5 minutes with engine stuff enabled

native kiln Jan 23, 2019, 8:41 PM

#

@glossy inlet ffs

#

a friend of me spammed me that link a dozen times because hes so hyped about it

#

and now you post it too

#

REEE

glossy inlet Jan 23, 2019, 8:41 PM

#

which link? 😄

native kiln Jan 23, 2019, 8:42 PM

#

the pr0gramm link

#

he even used it as his avatar https://s3.eu-central-1.amazonaws.com/trski-storage-public/ShareX/2019/01/2019-01-23_ivlRhaRcuoTTJqEbMi6.png but ill shall stop, too offtopic 😄

glossy inlet Jan 23, 2019, 8:44 PM

#

Profiler build is out
https://github.com/dedmen/ArmaScriptProfiler/releases/tag/1.0.28
with professional wiki page https://github.com/dedmen/ArmaScriptProfiler/wiki

#

Now It's time to lean back and do nothing for a couple days 🍪

smoky halo Jan 23, 2019, 9:04 PM

#

I wont look at it until tomorrow at the earliest so no bug reports rom me today at least.

glossy inlet Jan 23, 2019, 9:15 PM

#

I sure hope so.. I wanna go to bed and sleep well now..
Don't give me nightmares now

glossy inlet Jan 24, 2019, 1:04 PM

#

First performance optimizations being done using this shiny new thing

You can let the live recording run in the background. While you compare a old saved recording with the live one.
Red is old, yellow is new.
500us average down to 161us

📎 Tracy_2019-01-24_14-04-03.png

glossy inlet Jan 24, 2019, 2:40 PM

#

Next feature, capturing diag_log messages and displaying them in the trace.

Can just click onto the log message to jump to it in the trace.

📎 Tracy_2019-01-24_15-40-20.png

#

Don't know where a certain very cryptic RPT message comes from? if it comes from a script then the profiler will show you from where. (assuming the script itself also shows up in the profiler..)

cinder meteor Jan 24, 2019, 6:09 PM

#

I'm looking at Tracy...

void SetThreadName( std::thread& thread, const char* name );
void SetThreadName( std::thread::native_handle_type handle, const char* name );

I think we can't make it draw fake threads per every spawned .sqf script then, because Tracy is supposed to be attached to a real Windows thread... Am I right? :\

#

Unless we make windows create a thread, and then it will be dumping logged data into Tracy file

fallen stone Jan 24, 2019, 6:10 PM

#

Why would you want to?

#

Scheduled scripts are still executed linearly in one thread

#

Showing it as multiple would be confusing

glossy inlet Jan 24, 2019, 6:10 PM

#

We can do everything

#

it's programming

cinder meteor Jan 24, 2019, 6:11 PM

#

Yes we discussed it above... it would help me visualize states of my multiple spawned scripts with the help of tracy functionality, although they are inside one thread

fallen stone Jan 24, 2019, 6:12 PM

#

You'll have a lot of rows 😆

cinder meteor Jan 24, 2019, 6:12 PM

#

And?

fallen stone Jan 24, 2019, 6:12 PM

#

And nothing

cinder meteor Jan 24, 2019, 6:16 PM

#

Allright, let me explain @fallen stone
I have this huge CTI thing I am making... it has a 'thread' for handling AI, another thread for handling spawning/despawning things, there is also garrison level AI, group level AI, and unit level AI, groupped into different threads... I want to see better what's each thread is doing, even if it's essentially one thread. It'll just be easier for me to see debug info in a time plot than in an .rpt file, at least I hope so. Does it still make no sense in your opinion?

fallen stone Jan 24, 2019, 6:17 PM

#

To me, yes, it's going to be hard to have a proper timeline of events if you have god knows how many rows

#

Unless Tracy has a neat way to filter things

cinder meteor Jan 24, 2019, 6:18 PM

#

Hmm... no, I have like 3 'threads' actually, not so many

glossy inlet Jan 24, 2019, 6:18 PM

#

I can probably detect when a VM ends, and just re-use the threadid then

cinder meteor Jan 24, 2019, 6:18 PM

#

And I don't think you generally have hundreds of threads in a generic program... might be wrong, i'm not sure

fallen stone Jan 24, 2019, 6:19 PM

#

No, but you have a lot of scheduled things running over time

glossy inlet Jan 24, 2019, 6:20 PM

#

The number of threads will only be the max number of scheduled scripts that are running "in parallel"

#

which could be many.. But in that case you know which mod to remove, or where you have to fix your own shitcode

cinder meteor Jan 24, 2019, 6:20 PM

#

No, really I have a few 'threads'(below 5)... I don't spawn a scheduled script every time I want to do something 😄

fallen stone Jan 24, 2019, 6:21 PM

#

Your code isn't the only code running though

cinder meteor Jan 24, 2019, 6:21 PM

#

But let me correct myself, i was really talking about exposing tracy API for drawing stuff to SQF in Dedmen's addon, not making Dedmen's addon draw every spawned script by default

#

So I'd add tracy calls into my sqf scripts... and help me visualize stuff with it

#

where I need it

glossy inlet Jan 24, 2019, 6:22 PM

#

I could disable automatic scopes for scheduled, but leave the scope script commands open

fallen stone Jan 24, 2019, 6:22 PM

#

That would be better by the sounds

glossy inlet Jan 24, 2019, 6:23 PM

#

Would be cool if the profiler could define global macros so that you don't need to manually enable/disable the stuff

cinder meteor Jan 24, 2019, 6:24 PM

#

That's what I call a lot of rows 😄 (not my screenshot)
https://www.microsemi.com/images/soc/polarfire/Modelsim.png
These profilers remind me of my Verilog attempts so much

#

Hmm... you are talking about defining a macros in SQF file? like you can do with a C compiler...

glossy inlet Jan 24, 2019, 6:28 PM

#

The global macros in the compiler settings yeah.
My debugger had that for a while

glossy inlet Jan 25, 2019, 1:50 PM

#

[100%] Linking CXX shared library ArmaScriptProfiler.so
[100%] Built target ArmaScriptProfiler

Oh dis gon be gud

glossy inlet Jan 25, 2019, 3:38 PM

#

https://s.sqf.ovh/putty_2019-01-25_16-38-31.png Linux server profiling is a go! Even found a linux bug in Intercept 😄
Though who could expect that BI disables useful optimizations on Linux...

smoky halo Jan 25, 2019, 3:56 PM

#

Disables or not enables?

glossy inlet Jan 25, 2019, 4:21 PM

#

not enables I guess.. dunno

#

Windows version came first.. So.. Disables? 😄

sick verge Jan 25, 2019, 4:57 PM

#

@glossy inlet is that an real server monitor/profiler or is it your script profiler? 🤔

glossy inlet Jan 25, 2019, 5:06 PM

#

the script profiler

sick verge Jan 25, 2019, 5:16 PM

#

k

dawn palm Jan 30, 2019, 2:06 AM

#

subscriber only updates:
pboProject
deRtm
extractPbo (gui)

spice hawk Jan 31, 2019, 7:41 AM

#

How to make .wrp binarized ?

dawn palm Jan 31, 2019, 7:51 AM

#

pboProject uses bis binarise to achieve that miracle.

spice hawk Jan 31, 2019, 7:54 AM

#

Should I make any adjustments on pboProject @dawn palm

#

http://prntscr.com/mem3px

#

http://prntscr.com/mem3tj

dawn palm Jan 31, 2019, 8:54 AM

#

best off continuing in the terrain makers channel @spice hawk , they can help you far better than me, BUT
+your panels look fine
+your free version is way way out of date, a free update is available

spice hawk Jan 31, 2019, 10:18 AM

#

ok tahank you 4 answer @dawn palm

small sluice Feb 3, 2019, 10:19 AM

#

Does anyone have a copy of (or mirror for) the T_D tools?

vague shard Feb 3, 2019, 11:20 AM

#

@small sluice what are you looking for/want to do?

glossy inlet Feb 3, 2019, 12:24 PM

#

@small sluice #rules no crossposting

small sluice Feb 3, 2019, 3:06 PM

#

I want to see the named selections and texture files used in a model

sly skiff Feb 3, 2019, 3:12 PM

#

@small sluice https://community.bistudio.com/wiki/selectionNames

selectionNames

#

https://community.bistudio.com/wiki/getObjectTextures

getObjectTextures

small sluice Feb 3, 2019, 3:13 PM

#

Thanks, are they an object I can get the position of?

sly skiff Feb 3, 2019, 3:13 PM

#

well in a way but the selection position would be the average center of all the verticles in selection

#

why do you need them

#

if you are doing retextures you can only do them with things that have hidden selections configured into them

#

those you can also see from the config viewer

#

@small sluice

small sluice Feb 3, 2019, 3:19 PM

#

Thanks

silver carbon Feb 4, 2019, 12:43 AM

#

hey so when I go to publish my mod

https://i.imgur.com/U10yktq.png

Selecting this..

#

https://i.imgur.com/ZA9wOPn.png

Successfully published

#

it always defaults back to 'Friends' I have no idea why and I can't select public in workshop either

scenic canopy Feb 4, 2019, 1:08 PM

#

@silver carbon have you agreed to the latest workshop terms & licenses? I've had to fix that sometimes before we could use the workshop publishing stuff correctly

#

it's a small hidden banner when you visit your published workshop items, or something like that

dusky dune Feb 4, 2019, 2:50 PM

#

No small banners anywywhere on workshop

#

The account also has €5 in the wallet

#

(Its not marked as limited account)

silver carbon Feb 4, 2019, 2:55 PM

#

What adanteh said

dusky dune Feb 4, 2019, 2:58 PM

#

Turning on Steam guard did it (Whicvh is annoying as fug because we share the account, seeing workshop doesn't support multiple authors publishing updates)

#

Ofcourse there is absolutely no indication on this anywhere and steam's support page on limited accounts doesnt show it either

glossy inlet Feb 6, 2019, 3:11 PM

#

You always wanted to see how it looks like when a client joins your linux dedicated server and starts a mission? Well you are in luck!
(Linux server engine profiling support for profiler)

📎 Tracy_2019-02-06_16-10-34.png

#

Disabling just two zones (evwfGet, evGet) reduces the capture size from 1,8GB to about 400MB
The zone seem kinda useless for profiling and got like 10 million call count :U

#

Oh this is useful...
In general after disabling the spammy and mostly useless scopes, engine profiling becomes alot more useful as the traffic it generates isn't that big anymore.

📎 Tracy_2019-02-06_16-30-18.png

scenic canopy Feb 6, 2019, 4:14 PM

#

really cool!

glossy inlet Feb 6, 2019, 4:18 PM

#

CBA statemachine is wasting 1/4th of frametime doing nothing 🤔
I guess I broke something 😄

glossy inlet Feb 7, 2019, 9:36 AM

#

Well... Googling around, working on profiler, reading stackoverflow.. Ah a reply by Suma.. Wait.. Wat. Oh. Hi Suma.
https://stackoverflow.com/a/173925
😬
That guy is everywhere

smoky halo Feb 7, 2019, 9:50 AM

#

He's alive! https://stackoverflow.com/a/54207158/8746007

glossy inlet Feb 7, 2019, 10:38 AM

#

Which functions of your huge mod take the longest to compile? I have the answer!
(Special handling of calls to CBA_fnc_compileFunction)

📎 Tracy_2019-02-07_11-37-50.png

nocturne basin Feb 7, 2019, 10:50 AM

#

at least that one actually looks more like "questions nobody ever asked" to me

glossy inlet Feb 8, 2019, 3:45 PM

#

With the help of the profiler and 8 hours of work. I just about halved the mission preInit time with my modset.

With my ACE/CBA pull requests everyone will profit off of that in a couple weeks. With just ACE/CBA you can expect 1 to 2 seconds faster preInit :3

smoky halo Feb 8, 2019, 3:49 PM

#

So what was slowing it down?

smoky halo Feb 8, 2019, 4:11 PM

#

Did you manage to work out how to improve the per frame cost of the event handling in CBA?

glossy inlet Feb 8, 2019, 4:28 PM

#

There were functions compiled at preInit that should instead be compiled at preStart. Or config parsing done at preInit that should be at preStart.
Or my last change the eventhandlers are defined in config, and many are just
call compile preprocessFileLineNumbers "file"
So when it fires that event and executes that code, instead of just executing file it would always preproc and recompile the file before the actual EH code runs.
I added detection for that pattern and made it directly redirect to file when the eventhandlers are compiled (which I also moved from preInit to preStart)

So TLDR moving stuff from preInit to gamestart, and removing a stupid nested compile.

smoky halo Feb 8, 2019, 4:40 PM

#

So you made game start slower?

glossy inlet Feb 8, 2019, 4:42 PM

#

yeah. Game start slower once, instead of every mission start slow, besides the preprocess thing, that is just generally faster, not transferred to preStart

neon flax Feb 8, 2019, 4:43 PM

#

If you had world loading enabled at game start (menu bg world) then it should be same. Shouldnt it?

#

Because it was compiling the stuff anyway just not caching it.

#

Or is menu bg mission not firing preInit stuff?

glossy inlet Feb 8, 2019, 4:44 PM

#

It would cache to missionNamespace, which is reset before next preInit. So that wouldn't help much

#

https://s.sqf.ovh/Tracy_2019-02-08_17-46-48.png everything that lights up besides cba_ui_init is just gone after my fixes are merged.
CBA_ui_init stuff is moved to preStart
70% of cba common xeh_preInit is moved to preStart
500ms of ASR AI stuff is moved to preStart (not displayed on here)

https://github.com/CBATeam/CBA_A3/pull/1056#issuecomment-461842676 here is a summary of everything I did

#

The beautiful fixes are things like these:
https://github.com/acemod/ACEX/pull/169
https://github.com/acemod/ACE3/pull/6799

Stuff that has been in for 3 years and noone ever noticed that it's wrong.

glossy inlet Feb 8, 2019, 5:38 PM

#

@smoky halo can you explain? Do you mean the custom eventhandlers that CBA add's Like weapon changed and such?
CBA eventhandling itself has no per frame cost

smoky halo Feb 8, 2019, 5:38 PM

#

I seem to recall you saying that the per frame cost of CBA by default was quite high a week ago or so

glossy inlet Feb 8, 2019, 5:40 PM

#

because the player EH's are added always if you run ace I guess
a week ago definitely not, atleast I don't remember it

sacred flume Feb 9, 2019, 8:03 AM

#

Dedmen, for the interceptdb do connections persist on mission end in dedicated environment where server isn't restarting arma? Or does it close down connections on mission end and start back up with mission start?

sacred flume Feb 9, 2019, 9:50 AM

#

and I'm wondering how to validate connection is still active, I like to verify connection status regularly and flip a flag if connection to DB is lost for some reason so we no longer run DB calls

#

and do I use dbExecute instead of dbExecuteAsync to wait for query to execute before continuing? Or does dbExecuteAsync do that normally?

#

@glossy inlet

glossy inlet Feb 9, 2019, 11:11 AM

#

Connection should end when the SQF variable that stored the connection goes out of scope. So they should end at mission end. Unless you store the variable in uiNamespace or similar.

Connection is closed when there are no more references to the connection. If you have async threads running then the connection might only close 60 seconds after mission end, as the worker threads disconnect themselves 60 seconds after their last task.

#

dbExecute waits for the answer yes. dbExecuteAsync returns a "async result" that you can either wait on using dbWaitForResult or you can bind a callback to it dbBindCallback [{params ["_result","_arguments"]}, arguments]

sacred flume Feb 9, 2019, 11:25 AM

#

cool. Is there any way to validate connection? say the DB goes down and all the queries start failing?

glossy inlet Feb 9, 2019, 11:27 AM

#

If you str _connection it tells you whether it's connected

sacred flume Feb 9, 2019, 11:27 AM

#

ahh, cool

glossy inlet Feb 9, 2019, 11:28 AM

#

Either
<no session> or <not connected> or <connected to database: xxxx>
although no session should never happen 🤔

sacred flume Feb 9, 2019, 1:02 PM

#

does the mod file have to be in the root arma3 directory?

#

i suppose it does mmm

glossy inlet Feb 9, 2019, 1:04 PM

#

no

#

Same as any other mod

sacred flume Feb 9, 2019, 1:04 PM

#

i am getting a file not found error

glossy inlet Feb 9, 2019, 1:04 PM

#

for which file?

sacred flume Feb 9, 2019, 1:05 PM

#

"Database config error File not found: The operation completed successfully.: "#InterceptDB/config.yaml"
is the first line on loading up editor

#

yea, moving it locally fixed that

#

well, i managed to crash arma

#

lewl

glossy inlet Feb 9, 2019, 1:08 PM

#

https://github.com/intercept/intercept-database/blob/master/src/config.cpp#L29
Should be @InterceptDB mod folder in Arma directory. Just like every other mod... Unless you have a mods subfolder I guess

#

If you have a mdmp put it in a zip archive and throw it at me

#

If you don't use the config you can ignore that error btw

sacred flume Feb 9, 2019, 1:11 PM

#

lemme check to see if im doing something stupid first

glossy inlet Feb 9, 2019, 1:21 PM

#

Well my stuff should be stupidproof too

sacred flume Feb 9, 2019, 1:22 PM

#

mmm

#

i might just be totally misusing it lewl

#

im just checking where it f's up

#

DB_Connection = dbCreateConnection "maindb";
DB_Connection dbExecute (dbPrepareQueryConfig["newgame", [worldName]]);

That should be correct yes?

glossy inlet Feb 9, 2019, 1:27 PM

#

yeah. Don't see anything wrong. Maybe there is something wrong with the config? but then it should tell you and not crash

sacred flume Feb 9, 2019, 1:29 PM

#

yea, seems to crash on running DB_Connection dbExecute (dbPrepareQueryConfig["newgame", [worldName]]);

#

lemme pack

glossy inlet Feb 9, 2019, 1:33 PM

#

A breakpoint was encountered. whoops 😄

#

https://github.com/intercept/intercept-database/blob/5b5eacc286eed729cdfd1b880335a22d969b1107/src/connection.cpp#L193 there it is.
Question is why that fails 🤔 I'll add logging, gimme a minute

sacred flume Feb 9, 2019, 1:34 PM

#

lol

glossy inlet Feb 9, 2019, 1:49 PM

#

Man that takes a long time to build

📎 intercept-database_x64.dll

#

should print to RPT now on error

sacred flume Feb 9, 2019, 1:54 PM

#

ahhhh, ok i see error now

#

auth level on db user, forgot to add execute

#

does dbCreateConnection not wait for connect to complete or fail?

glossy inlet Feb 9, 2019, 2:09 PM

#

no

#

connection is only established at first query

sacred flume Feb 9, 2019, 2:10 PM

#

ahhh, ok

#

if first query attempt fails to connect, will it throw error and break script?

glossy inlet Feb 9, 2019, 2:15 PM

#

it will throw error to RPT and return a empty query result

#

no script error no

sacred flume Feb 9, 2019, 2:15 PM

#

kk

#

yay, its connecting

#

i do have the problem of it always saying <not connected> when i check it

#

but it does apparently connect on queries

#

so not something I can use to validate if connection succeeded

#

i guess i'll just have to validate every return and use that

glossy inlet Feb 9, 2019, 2:45 PM

#

oh.. Right. Because it creates a copy of the connection in the worker thread

#

You could just make a ping query I guess ^^

sacred flume Feb 9, 2019, 2:47 PM

#

yea pretty much what ill do

glossy inlet Feb 9, 2019, 2:47 PM

#

I'll come up with a fix for that when I have time to work on it.
Currently working on making missions load 8 seconds faster :U

sacred flume Feb 9, 2019, 2:47 PM

#

i have a loop for server details every so often, ill check there so silent errors wont last long

glossy inlet Feb 9, 2019, 2:48 PM

#

I could build something like.. You define eventhandlers on a connection. And when an error happens the eventhandler will be called. That would be good right?

#

Feel free to make feature requests on the github with whatever comes to your mind
https://github.com/intercept/intercept-database/issues

sacred flume Feb 9, 2019, 2:49 PM

#

i think so if i understand right, i just want to flip a flag I use before entering DB related code

#

set it to false, so DB code doesnt get entered

#

no point in any loops or pulling information to send to DB if DB has no connection

glossy inlet Feb 9, 2019, 2:51 PM

#

I could make a ping command that makes sure it connects and that it can query. Basically shortcut for SELECT 1 and checking if it returns 1

sacred flume Feb 9, 2019, 2:52 PM

#

yep

#

would work for me

#

moving this mission over today, test through tomorrow, maybe have live for this week

#

see how it runs through extended use

glossy inlet Feb 9, 2019, 5:10 PM

#

The profiler just paid for itself. Two days of work + profiler.

Before is CBA and ACE on latest release, plus my default modset (RHS, CUP Terrains, NiArms, 3CB and some misc stuff).
After is CBA and ACE on master branch and all my PR's merged.

Before:
preStart: 21.92s
preInit: 7.77s
postInit: 11.61s

After:
preStart: 32.04s
preInit: 2.52s
postInit: 1.84s

Totally worth it.

>15 seconds faster mission start for 11 seconds longer game start

only ace,cba

Before
preStart: 15.72s
preInit: 4.06s
postInit: 5.10s

After
preStart: 19.89s
preInit: 1.36s
postInit: 1.19s

sick verge Feb 9, 2019, 5:17 PM

#

👌

tight cypress Feb 14, 2019, 10:49 AM

#

Great job!

vague shard Feb 15, 2019, 3:29 PM

#

is it possible to upload arma mods to workshop without steam running?

#

or stop steam downloading anything from workshop

cobalt stag Feb 15, 2019, 3:31 PM

#

@vague shard just set a download speed to possible minimum, upload what you want, and remove the Steam

#

In my case - my mods installed on a main PC are downloading already a few weeks in the line on my second PC. I don't need mods there currently.

#

🙂

vague shard Feb 15, 2019, 3:37 PM

#

speed limiter is an interesting idea indeed

scenic canopy Feb 15, 2019, 3:41 PM

#

@vague shard you can use steamcmd to upload without having steam installed or running