code-talk

[fickle aurora]

the reason why buffer overflow is a problem is because the CPU has to look in subsequent memory locations until it reads null

[devout ivy]

Almost as if null terminated strings are a problem

And as if intel provide instructions for prefixed strings for a valid reason

[fickle aurora]

yes it is a problem only if you don't know what you are doing

[long raft]

same problem even with length prefix, but maybe less likely

yea allan this is ridiculous, youre basically giving a skill issue git gud argument to programming

[devout ivy]

Typically can test beforehand with a prefixed string

[fickle aurora]

derp is talking about one single value, which isn't a problem because it will only access the maximum number of memory locations for whatever data type it is

[long raft]

why have safety features on a car if you can just drive better

[devout ivy]

You've completely missed derps point

In a rather hilariously inept manner

[long raft]

you were suggesting one single value could overwrite your bootloader. im suggesting a single mathematical operation thats done incorrectly is a big deal

[fickle aurora]

I never missed the point, adding 1 to 255 in a single byte won't cause an issue

[devout ivy]

Are you sure? It's not a memory violation sure

[long raft]

is the difference of looping 256 times and 0

[devout ivy]

But it could still be a nasty bug to your programs logic

[fickle aurora]

it's not a memory violation because a byte declaration will limit the CPU to only read one byte of memory

[long raft]

yea this isnt mangling a string, this is causing logic errors

[devout ivy]

I said it wasn't a memory violation

[long raft]

and buffer overrun testing is comparable to integer overflows was my point

[devout ivy]

You don't need to explain why it wasn't a memory violation

[fickle aurora]

logic errors aren't even an issue because it will just set some flags in the flags register, and most people never even check eflags

[devout ivy]

You've just been fired

[long raft]

..

[devout ivy]

Logic errors can and have been life or death

[long raft]

i do not believe allan is trolling

[fickle aurora]

you are confusing a data type with an array of data types

[devout ivy]

No

I'm not

So if I am writing safety critical code

And a 255 incorrectly becomes a 0

[long raft]

arithmetic errors are very grievous

[devout ivy]

And results in some other safety critical code not running as the condition is now wrong

That's not a problem?

[long raft]

they are easier to debug than a buffer overrun tho, oof

just step through a scenario that fails

[fickle aurora]

how is buffer overflow not easy to debug?

[long raft]

if there was some easy way to say why, then it would be easy

[devout ivy]

I'm guessing you don't believe in unit testing either

[long raft]

i would say the biggest reason is because you cant easily visualize the data youre reading, its not like a string. you can hex edit it, but encoded things are difficult to analyze

oh shit test driven development is 21st century, just git gud

we didnt need tests in the 20th century

there was literally never a problem until it rolled over into the 21st century, very first day even

err first day of 21st century is 2001 huh

[fickle aurora]

I'm pretty sure people were better at debugging in the 90s than now

[long raft]

hah.

[fickle aurora]

I can use star citizen as an example of a project where code is pumped out with not even one glance at bugs

[devout ivy]

That doesn't answer the question

[long raft]

its probably 5 orders of magnitude more complex

you already know your answer six

he doesnt approve of buffer overrun testing

you think he wants unit tests?

[fickle aurora]

in the 90's, if your program was bugged out, people didn't buy it because you couldn't just automatically patch it by connecting to the internet

[long raft]

unit tests would just be admitting youre fallible

[devout ivy]

Clearly has no experience in embedded design then

[long raft]

oh well i dont either

[devout ivy]

Or even touched basics of safety critical programming

[fickle aurora]

well for one, no one uses C# in embedded

[long raft]

...

[devout ivy]

Because that cavalier attitude has literally resulted in deaths in medical and industrial sectors

[long raft]

no one?

[devout ivy]

Did I say they did

And actually, they do

[long raft]

is it rare?

[devout ivy]

Extraordinarily rare yes

But not non existent

[fickle aurora]

it is non existent if you are referring to microcontrollers

[devout ivy]

So you're saying I don't have an esp32 this minute running c#

[fickle aurora]

esp32 is a microprocessor, not a microcontroller

[devout ivy]

It is categorised by it's own vendor as a microcontroller

[long raft]

https://tenor.com/view/andy-davidson-andy-torchwood-confused-meme-meme-gif-20411968

[devout ivy]

I also have stm32's running it

Which you yourself already referred to as a microcontroller

[fickle aurora]

yes they like to say it's that, but it's Tensilica Xtensa LX6 microprocessor @ 160 or 240 MHz

[devout ivy]

It has no mmu

[long raft]

i feel like if i came in here as an alt talking about how great C is allan would find a reason C sucks

[fickle aurora]

microcontrollers don't run at high clock speeds like that either

[long raft]

especially if i had a blue name...

[devout ivy]

You've clearly not touched any microcontroller in the last decade then

We even have 1ghz microcontrollers today

[long raft]

i guess every programmer needs a creed, i dunno i guess allans is that things peaked in 1970

[devout ivy]

Mmu-less as that seems to result in the common modern distinction

Even atmel avr32's are clocked that fast (hundreds of megs not gig)

[long raft]

imagine doing async in C

[fickle aurora]

it's hardly believable that anyone has used C# for a microcontroller

[devout ivy]

C# can run on an arm cortex m0 mcu

There's been several ways to do so over the years

Starting with the .net compact framework

.net microframework

Nanoframework

[long raft]

i wonder what advantage there could be to doing that even

[devout ivy]

And also meadow

[fickle aurora]

arm is a microprocessor.. again, not a microcontroller

[devout ivy]

Cortex M ksnt

[long raft]

all i can think of is like ... if you need a fast web server? i dunno

[devout ivy]

Not cortex A, cortex M

[long raft]

if you just really only know 1 language i guess

[devout ivy]

Cortex M is the microcontroller variant

To compete with avr and pic offerings

They lack MMUs, microprocessors have MMUs

[long raft]

uwu?

[devout ivy]

Or is the humble stm32 not a microcontroller now

[fickle aurora]

yes I am sure someone managed to fit the .net framework into it, but it's not mainstream and it's not the intended method of programming a microcontroller

[devout ivy]

Change in tune from earlier where you claimed it couldn't happen

[fickle aurora]

here is a good reference to your question/suggestion https://electronics.stackexchange.com/questions/284715/write-program-for-atmega-microcontroller-with-c

[devout ivy]

What question or suggestion

I made no question or suggestion

I just made the objective statement that c# can run on several mcus

[fickle aurora]

well, the answer is there is a project called netduino for getting .net into microcontrollers

[devout ivy]

I didn't say it ran on all of them

Netduino died years ago

[long raft]

oh netduino is old - even ive heard of that

[devout ivy]

Replaced with nanoframework and meadow

[long raft]

i didnt know an arduino counted as a microwhatever

[devout ivy]

Still never answered my actual question

[fickle aurora]

well you guys keep trolling with obvious stuff, like with buffer overflows going on about storing value 256 in a one byte memory location, which wouldn't do anything of course because you can only fit 8 bits in one address, and of course the real issue with buffer overrun is quite clearly a problem with pointers to arrays

[devout ivy]

No, that was you missing the point

[fickle aurora]

at first I couldn't tell if you really didn't understand but I kind of figured it out by now

[devout ivy]

Nor was it trolling

If anyone doesn't understand, it's you. You don't understand that storing 256 in a 1 byte address is a logical error

Which may set a flag. If you bother to read that flag

What if you dont

What if that logical error results in other safety critical code failing

[fickle aurora]

it isn't even an error, it may be a design problem but it isn't a security violation

[devout ivy]

Didn't say it was

[fickle aurora]

pointers to arrays are an actual issue

[devout ivy]

So a logical error in a safety critical system like I don't know, a pacemaker, that isn't a problem

A bug in an insulin pump firmware from someone not checking integer overflow, not unheard of

[long raft]

what is the difference of a buffer overrun and a [bad] pointer? its the same thing in reverse. its a buffer underrun or a combination of both

[devout ivy]

And your attitude seems to be that it's not important. And seemingly that unit tests to verify this kind of thing are also not important

And that tools that exist today for reducing errors as simple as integer under/overflow or out of bounds memory issues are also pointless?

[long raft]

AllanOS would release 1.0 on schedule, and never need another update again

[fickle aurora]

it's a design problem, you can sort it out by reading the eflags register, it's not a problem that could cause havoc on a computer system

[long raft]

just git good lol

just point your pointers to the right place, whats the problem

[devout ivy]

And that attitude just got you fired again from a place I previously worked

It might not cause havoc as far as the computer itself cares. But if that bug results in your ABS system not responding correctly?

You still seem allergic to even mentioning the words unit test

[long raft]

hasnt got to that chapter in the old book on C

its probably ... past the index

maybe an addendum...

[devout ivy]

Eh, it's a principle that existed as far back as the 80s

[long raft]

yes i think in the modern day, we dont expect programmers to be perfect, we know things are complicated, so we pay some CPU time to double check their work, and its worth it because we have CPU time in spades

[devout ivy]

And tooling did exist back that far

[long raft]

it existed in manufacturing, not in computers software

cmon, 700 line programs dont need unit testing. they are the unit test

[devout ivy]

Nah, it just wasn't dubbed unit testing til the late 90s

But testing code via execution in smaller chunks was performed even in early 80s

[long raft]

yea, and the code bases were small

[devout ivy]

Even 8bit avr code can be unit tested. Assembly is harder to unit test, it's one of the major reasons you see less of it around

But it can be done, and it is actively done

[long raft]

eh its easy you just step through it, and then never ever ever change it again

you cant refactor assembly. all you can do is rewrite it, if its been more than like 2 days

[fickle aurora]

since when was assembly harder to test?

can't refactor? since when?

so you're saying once I write a program in assembly, it's not possible to modify the program?

[devout ivy]

unit test. And note I said it can and is done

[long raft]

assembly is largely unreadable

its hard to rewrite stuff better

[fickle aurora]

jeez I must be doing things the wrong way, because I have no problem changing or updating assembly

[long raft]

i mean you cant make it more readable because its ...inherently unreadable

[devout ivy]

Eh, I've seen some clean ASM, I've seen obscure ASM, I've seen dirty ASM

It can be read

[fickle aurora]

it's not unreadable if you learn it

[long raft]

3 comments for each line?

[devout ivy]

Don't need that usually

But typically assembly isn't as broken down into smaller discrete units

Can't unit test something that isn't a unit

[long raft]

patterns are harder to recognize

[devout ivy]

But it can 100% be done

[fickle aurora]

you absolutely can break it down into smaller sections

[devout ivy]

Where did I say it couldn't be

[long raft]

yea i guess my point was its easier to rewrite it all than refactor it

[fickle aurora]

you can include other files, libraries, execute functions written in other languages

[devout ivy]

I can't say I actually agree on that one. Really does come down to having learned the bastard thing properly

There will be obscurities in any language. Comment lines exist for that

[long raft]

what is there to learn? its mostly referencing

its not like theres tricks to assembly

[devout ivy]

There are, some are simple and some are almost galaxy brain

[long raft]

really?

galaxy brain assembly tricks?

[devout ivy]

Everyone knows about the subtle shift by 1 space instead of mul/div by 2

[long raft]

errr i ... know of that myth yes

[devout ivy]

Or testing for odds/evens by testing the LSB

[fickle aurora]

you only need to learn the 20 most commonly used instructions for x64 to make basic functions

and the calling conventions are important

[long raft]

the conventions are very simple compared to any other language

[fickle aurora]

GCC ignores some of the calling conventions though

[devout ivy]

It's not a myth but it isn't as big as impact today as it used to be as modern mul/div is significantly faster than it was and most compilers are smart enough to make the same optimisation

[fickle aurora]

GCC won't add 32 bytes of reserved stack space for each function call as directed by Microsoft, but it's not critical

[long raft]

yes these days its not faster for left shift, and right shift means you wont also have a remainder to use

[devout ivy]

Muls in particular are very fast on modern CPUs yes

[long raft]

yea those ... microoptimizations, like xor to zero a register, i never considered those galaxy brain, those are just architecture quirks, reference things

[devout ivy]

I think single cycle now in some arm core designs

[fickle aurora]

the biggest advantage is being able to write a program in as few instructions as possible

[long raft]

why is that an advantage?

what if theyre slower instructions

[fickle aurora]

even just a hello world program you can cut the bloatware by at least 50% by writing the assembly yourself instead of using GCC

[devout ivy]

Div is still slower. And on mcu's you're less likely to find cutting edge stuff like a single cycle mul instruction

[long raft]

i have a big hard drive and lots of ram, why do i care about that

[devout ivy]

Now go and cram some code in literally 32 bytes of ram

[fickle aurora]

if you ran concurrent servers with millions of connections you might care about it

[long raft]

its not an issue

[devout ivy]

Humble Arduino Uno uses an atmega 328p

8bit avr chipset at 16mhz

Like 2kb of ram iirc

[long raft]

hmm not an issue for me

[devout ivy]

There's the sort of place it might legitimately make a difference

[long raft]

also i still think a C/c++ compiler would do a better job optimizing size than i could do, or anyone else i know

[devout ivy]

Wouldn't want to go and write an entire computer game on a modern pc in it, and most game engines today don't have any assembly at all anymore

Not even 1 little sub routine called from c

[fickle aurora]

Factorio was written in C on allegro originally

not a very old game

[devout ivy]

Yes, C

Not assembly

I didn't say nothing was written in C

Albeit checked, it was C++

And still uses a lot of lua

[fickle aurora]

derp, a 1 hour MIT free open courseware (University of Massachusetts) will teach you how to make assembly more optimized than a C/C++ compiler

all the courses are available if you want to learn

of course, if you don't want to become better, that is also an option

[long raft]

i uhh ... i reject the notion that its possible

its sheer arrogance to think youre better than a compiler

its like a fundamental basis for my whole way of thinking. also there is no such thing as a free lunch

[devout ivy]

It's actually been regularly proven wrong too that the compiler is slower than a human. You absolutely can cherry pick examples like hello world

But real code isn't always hello world

[long raft]

i also cannot beat a computer at chess

[devout ivy]

Human written assembly is usually terrible for cache alignment, compilers generate for cache alignment, can be hugely important on hot paths

[long raft]

yea cache ... ffs... cache.

been doing a lot of GPU programming lately, omg nothing from the CPU world works

all that stuff you thought you learned was fast its not

[devout ivy]

Does it mean there are humans that can't write good code with excellent caching performance, no, of course some can

[long raft]

with tons and tons of testing

and deep understanding of the particular hardware architecture

which will be useless in 3 years

[devout ivy]

Fewer instructions doesn't always mean faster either as not all instructions are single cycle. Particularly in x86

Many arent

Throw in a bit of branch prediction and out of order execution to keep you on your toes - and compiler on its toes tbf

[long raft]

you have to think more stochastically at these complexity levels now

[devout ivy]

Arm typically has much lower cycle counts

[long raft]

how is RISC pushing so hard now? like 15 years ago they were totally hobbled by heat

[devout ivy]

They were never hobbled

The money just wasn't there

[long raft]

they werent seeing the performance gains of cisc

[devout ivy]

The only interest was more in how can we make tiny chip power tiny device rather than pushing the envelope of what 100W can do

[long raft]

apple abandoned the architecture at least in part because of heat issues for higher speeds and mobile devices was a no go

intel was even killing it on the low power stuff

now arm processors are so stupidly good, and m1

i think only one of them will survive into the future, i guess risc?

[devout ivy]

PowerPC was pretty fast. But it was partly business decisions, apple small customer and IBM not catering their lines to apple

Less a fundamental issue

[long raft]

really?

that seems like a petty reason to change your architecture every decade

its quite disruptive

[devout ivy]

They sold more 360's than apple sold PowerPC macs in total

360, GameCube, Wii, Wii u and PS3 all featured powerpc

[long raft]

oh man that 360 was a monster

so cheap for that kind of power. but it also melted

[devout ivy]

Yep and it's a risc design

[long raft]

rrodisc

[devout ivy]

Yeah but remember the first revision units, the coolers aren't torqued to the right spec

[long raft]

that console lasted like 5 years and was still relevant ... damn that was good hardware

[devout ivy]

So were usually making poor die contact

[long raft]

yea mine melted

[devout ivy]

Mine came to life in an oven

[long raft]

i had mine serviced, they extended the warranty an extra year and mine melted in that window

i still have it lol

[devout ivy]

But yeah basically, IBM weren't releasing new chips suitable as upgrades to what apple were selling

[long raft]

so i dunno yes the power was there, but also the heat was. because yes even when they replaced the cooling unit i could use that thing as a house heater

[devout ivy]

Like sure, they had a Mac on sale, but next gen has to be faster right? Part didn't exist in the catalogue

[fickle aurora]

https://www.youtube.com/watch?v=wt7a5BOztuM at 2:40 you can see where one of the examples of why you would want to learn assembly is due to compiler inefficiencies and problems

[long raft]

with apple? nah it doesnt have to be faster

[devout ivy]

Apple were less dickish back then

[long raft]

the only people that course should be relevant to are compiler designers

[devout ivy]

People were legitimately using macs for good editing performance back then

[fickle aurora]

no, it's for general programmers

[long raft]

i remember macs were stable for video editing, never performance, but you could not get stability for video in a windows program

video is ... so intense. good god is it intense.

[devout ivy]

When the first PowerPC macs came out. You were literally looking at gains over competition that make the M1 today look like a slouch versus it's rivals

And the crazy thing with PowerPC macs. Idle power draw. Full load they pulled many watts sure.

[long raft]

oh they can clock them way down for idle huh

[devout ivy]

But they had CPUs in 2000 with 1W idle powers

[long raft]

definitely not my xbox 360 lol

[devout ivy]

My complete Mac MDD with dual powerpc's idles at 20W from wall. Less than my lattepanda delta pulls at idle

And that's a modern atom (branded as celeron)

[long raft]

atoms were disappointingly weak IMO

[devout ivy]

Modern atom isn't as bad tbf

[long raft]

its like the price you pay for x86

[devout ivy]

But the lattepanda is bad

It's just taken a long time for companies to pour the money into another personal computer take on risc unfortunately

And even that I think is just side effect of the rampant consumerism in modern flagship smartphones having to be more powerful than units a year prior

And finally just realising we can take that one extra step. And apples closed wall ecosystem giving them the weight to take that risk

Now I'm hoping the windows 10/Qualcomm exclusivity deal ends (meant to this coming spring) and that M1 lights a fire under competitors butts

[long raft]

wait what about windows 10 and qualcomm?

was there a windows 10 for ...snapdragons or something?

[devout ivy]

Windows 10 for arm

Only contains Snapdragon drivers

And the only windows 10 arm devices released are all Snapdragon

[long raft]

I had no idea they were exclusive

[devout ivy]

Even the surface pro X and it's "SQ1"/SQ2

Those are Qualcomm Snapdragon 8cx's with a neutral net co processor and a slight overclock

And the 8cx is just an 855 from phones with more cache, an overclock and some extra I/O

And then the original windows 10 arm devices are literally Snapdragon 835. Then some used the 845

Have a surface pro X at work, it's interesting

Nowhere near same performance League as m1

[fickle aurora]

https://www.youtube.com/watch?v=pv4Yq35Chx0

[chilly burrow]

  "warId" : "4d5467c6-3ac9-433e-ba0c-c95fd73916e9",
  "warNumber" : 86,
  "winner" : "NONE",
  "conquestStartTime" : 1639507210409,
  "conquestEndTime" : null,
  "resistanceStartTime" : null,
  "requiredVictoryTowns" : 32
}```
When a faction wins/loses the war, does the "warNumber" changes, or it stays the same until next war starts?

[devout ivy]

Believe same until next war

[hollow gyro]

I reckon so too, but I'm only recently started this kind of coding.
I reckon it stays that way until Resistance Phase ends.

[pure sierra]

correct, resistance phase keeps same war number

[fickle aurora]

https://pastebin.com/5GCkTDsR good simple example for managing custom memory blocks ... so instead of reallocating with realloc which could be inefficient, you could reserve a big block for your program and just re-write over all the memory locations however you like

[long raft]

what advantage is there to reinventing the OSs memory management?

its just another layer

if you want to avoid alloc calls then you would typically just use a memory pool, much easier to allocate smaller pieces than a big block

[fickle aurora]

I'm offering examples for those who like to learn more about it

and it seems you don't know what you are talking about, like yesterday when you mentioned doing Byte 255;, in C, there is no Byte, but you can use char.. except you can't store 255 in a char as the max is 127, now unsigned char can store up to 255 .. and the reason for this is that the most significant bit is reserved for the sign flag

[long raft]

who said that was C code?

it was just an illustration

typedef unsigned char byte;
now its C compatible

[coral sundial]

Why do you need to typedef defined types?

[long raft]

its a C tradition to make your code as unreadable as possible

[long raft]

has anyone stitched the newest map?

[pure sierra]

what maps

[long raft]

i know there are some road changes but im not seeing any updates...

[pure sierra]

i think i forgot

[long raft]

i forgor too

[pure sierra]

thanks for reminding me

[long raft]

but i see nothing has updated in the warapi so i dunno if anyone has stitched it from the assets yet

storm cannons should be coming online soon so thatll be in the warapi to finally test

id like a new map tho and i havent kept my code up to date for stitching it, and i dont have the assets from the game directly

[pure sierra]

the maps did update for last update in warapi

[long raft]

28 days ago?

[pure sierra]

well that had the update for the sc

[long raft]

Hmm OK I'll use that one

[coral sundial]

@long raft you and my c teacher would go along wel

[barren quarry]

Great march and heartlands

I think

I updated the tiles

[pure sierra]

storm cannons are alive

[long raft]

thanks, I think I'm gonna switch to your sat maps, they are hot and they show errors for roads easily.

[barren quarry]

sat maps were not updated

they are fucked

[long raft]

😦

damn they look good tho

damn i thought it was gonna be amazing for finding missed roads

[pure sierra]

height maps ?

[barren quarry]

the unprocessed tiles

what i mean by fucked

[lunar cobalt]

holy shit

[long raft]

uhh

what about the regions where the map changed?

[long raft]

is it possible only those regions are ok and the old ones can be used for the rest?

[barren quarry]

hmm i think you could be right

[barren quarry]

@pure sierra you need to specify which number belongs to which faction

[pure sierra]

if you hover it tells you, how else would you indicate without taking up extra width ?

[frosty sierra]

"C" or "W" infront of the number

Still takes space but not much

[pure sierra]

thats extra width i dont have

[pure sierra]

unless i make it a really small font

[barren quarry]

nop

okay

yea

[pure sierra]

i changed it

reload ?

[barren quarry]

yes reload worked

[lunar cobalt]

Color it to the factions

[vague otter]

Why have total cadualties that prominent anyhow? It's not that critical

You could have them as a hover over hourly cssualty rates for ecample

[mossy pelican]

i was about to say change the color of the font to be the color of the faction but icanari already said it

[pure sierra]

tried that but green text against green bg not great, you can get the team by hover

@vague otter ive been seriously considering this too

Iive moved the rocket back up for now

[vague otter]

Just out of curiosity, what was the reason to include it as it is?

[pure sierra]

Just to show it?

[pure sierra]

https://github.com/clapfoot/warapi/issues/98

[long raft]

@barren quarry do you have a zip of the new unprocessed map images and/or the old ones?

[barren quarry]

I will have to check if I have the old ones because I was cleaning my hard drive recently

New ones yes

But I'm not at my pc right now

[long raft]

ok np

[lunar cobalt]

I think I have the old ones

zipping them currently

[long raft]

https://tenor.com/view/chico-cute-tantrum-food-give-me-food-gif-16475884

[lunar cobalt]

@long raft

[long raft]

om nom nom ty

[lunar cobalt]

I think these are U46 or U46 devbranch

Not sure which

Probably U46

[long raft]

reaching trail is just a black image in that

[lunar cobalt]

correct

[long raft]

hmm

good god this file is so big i cant even open it. the webp is 47mb

where do i download more ram?

[pure sierra]

where ?

[coral sundial]

Downloadmoreram.com

[dark kayak]

Go to that one really sketchy site and click on every link then bam! You get like 75 viruses and maybe one of them will give you more ram

[mossy pelican]

If a virus infects your OS to the point that it no longer works then it technically gives you more RAM by freeing up the RAM your system is using

[coral sundial]

Dont the virussus(virussi) use ram

[mossy pelican]

depends on the virus

some might use disk speed

or be cpu intensive

[languid harness]

just delete system32

[long raft]

wow deleting system32 gave me a 10x speedup

[lunar cobalt]

nice

[tribal orchid]

Is there a way to find out the number of resources in a building?

I looked a bit but couldn't find it

[languid harness]

You mean via API?

[tribal orchid]

There is already an API, but there is no endpoint showing the number of BMATs in city x

[languid harness]

I know there is an api

What I asked if your question relates to the api or in game

And no, there is no end point in the api because its faction specific data

[tribal orchid]

sadge

[languid harness]

And the api contains no data that only one faction can see

[pure sierra]

well the number of ranges is getting a bit out of hand

[pure sierra]

https://www.reddit.com/r/foxholegame/comments/rp3nub/anyone_worked_out_a_way_to_change_mouse_cursor/

[frozen oriole]

Yep, the mod should he somewhere in the FMN discord

[languid harness]

Barely readable at this point goddamn

[vague otter]

hover is your saviour

display range on hover only, and optional toggle to display all ranges

[long raft]

i didnt realize the intel stations were part of the api too... not sure how to make use of those

[pure sierra]

might have to

[neat fossil]

so we talk in code here?

[devout ivy]

azetc bqeqf wqgpw qdolz qczis mgeow xrgup snujq ffnob ezzbp sekbh ivfr

[coral sundial]

Beezlebup

[hollow gyro]

Bah weep granah weep nini bong

[devout ivy]

Well, if it makes it any easier, I didn't use a plugboard

[languid harness]

Programming code

[pure sierra]

discount codes 🙂

[long raft]

ISBNs and only ISBNs

[plush dew]

Apha Mike foxtrot

[past crag]

Is there a comprehensive list of all the API calls that one can make documented anywhere?

[long raft]

on the github warapi

[past crag]

Is there any additional api queries that can be made such as unique spawns in a region etc?

[long raft]

yes, enlistments

its a value in one of the queries somewhere, you can find unique enlistments in a region since the war started (home regions are exceptional)

[hollow gyro]

GET /worldconquest/warReport/:mapName

{
  "totalEnlistments" : 148,
  "colonialCasualties" : 202,
  "wardenCasualties" : 222,
  "dayOfWar" : 2
}

[hollow gyro]

See the first pin

[pure sierra]

@past crag just in case its not clear, it is only unique per region not across maps

[past crag]

thanks!

[pure sierra]

well that escalated quickly

[long raft]

seems like a lot of bad sectors.

e.g. bad cable instead

[oblique crag]

12 days left 😮

[long raft]

is it SSD?

[pure sierra]

no

[oblique crag]

RAID 10 ?

[devout ivy]

19120 hours on spinning disk

[long raft]

how does SMART think it knows a magnetic drive has ... time left

magnetic drives are binary, they go from working to not working

[devout ivy]

prediction. Its taken 19120 hours for x sectors out of y to go bad, and minimum operating number of sectors is z, assume linear progression and z achieved in 12 days

Accuracy? minimal. But there'll be method in the madness

[long raft]

i know but why would you assume linear progression with a spinning disk

[oblique crag]

maybe too much shadowplay recording on that drive

[devout ivy]

Eh, they may try curve fitting or something

[long raft]

you know whats nuts about a drive? after they reach like 5 years their likelihood of death in a given year begins to drop

then again those are ~2010-2013 drives

[devout ivy]

tbf with consumer drives too. Even the orientation its mounted in makes a massive different to MTBF

[long raft]

ive never lost a single one of those 4T HGST drives before and ive had like ... 30-40 of them

[devout ivy]

flat as in a typical drive cage in a desktop tends to be fine

[pure sierra]

the seagates of this error seem to die quick

[long raft]

seagates of that era were awful

[devout ivy]

on side as in some SFF PC's and caddies, tends to murder consumer hard disks quickly

I'm yet to own a hitachi drive for more than 6 months at a time, stopped buying them entirely

[mossy pelican]

at least its not 420 bad sectors

[long raft]

hitachi is mediocre

[devout ivy]

and also stopped warranty claiming on the machine I had that came with one from factory after the 3rd drive

purchased a WD drive for that machine instead

[long raft]

seagate once again seems like the best drives now for the large 12T+ stuff

[pure sierra]

yes i have 2 seagotes of same era and age in similar condition now, my mate stopped buying them after than era

[devout ivy]

ordinarily not knowing the level of activity on the drive, I wouldnt really expect a failure at that age. But its definitely possible

[long raft]

8 years?

thats a good life

[devout ivy]

thats a lot less than 8 years 😛

[long raft]

hmm

[devout ivy]

2 and a bit

[long raft]

for a 2TB drive?

[oblique crag]

maybe 2nd hand drive

[devout ivy]

those times are taken from the drives controller

[long raft]

so run time, but not age

[devout ivy]

so would count the time in previous owners possession

[pure sierra]

not sure about that

[devout ivy]

Thats also fair, dunno if it counts up when head is parked

[pure sierra]

this is my other one

[long raft]

8 bad sectors o_O

i consider a drive failed if it has 1 bad sector

[devout ivy]

Likewise

[pure sierra]

oh so thats bad

[long raft]

i dunno. maybe my tolerance is low

[devout ivy]

Generally drives tend to be 0 failed sectors, then death

[pure sierra]

well that other one had 320 !

[devout ivy]

yeah, likely all in a short time frame

[pure sierra]

its in a windows raid

[long raft]

i dunno if thats data stored on the disk or the OS reporting bad sectors

because a bad sata cable will give you tons of bad sector reads

[devout ivy]

If its from SMART, it'll be controller side

[pure sierra]

havent touched that box for a number of days

[long raft]

if its SMART then id replace ... probably both drives

[pure sierra]

i wont trust them for anything but temp disposable storage

[devout ivy]

so long as you have backups of anything important

[long raft]

this reminds me to check my zpools lol

[oblique crag]

dont backup , live dangerously

[pure sierra]

i was running as raid for this reason.

[devout ivy]

raid isnt a backup

[pure sierra]

kinda of

better than nothing

[long raft]

i raid, i snapshot, and i have offsite backup

and im still worried

[devout ivy]

Barely. Raid gives you a chance of keeping the working copy only. But should never be trusted as a backup strategy

[long raft]

i used to have this NAS with an annoying red hard drive light and i electrical taped over it to sleep, and one day i realized both disks have failed and i lost everything for years

[devout ivy]

I do intend on building a NAS soon

my current backup strategy is frankly disorganised and poor XD

[pure sierra]

oof, yeah, that caught me too , windows raid does not give you any notifications if the raid is not working,

so what i did is put computer management in the startup, so each boot i check it manually

[long raft]

better than hardware raid

[devout ivy]

best part with hardware raid is what happened at work recently - thankfully in a scenario that had full backup&recovery plans so was mitigated quickly

Losing the raid controller itself

[long raft]

thats the #1 reason it sucks

[pure sierra]

yes that what scares me of hardware raid, not being able to buy another card

[devout ivy]

Spare cards were kept on hand for that very reason

[long raft]

there are other annoying parts of hardware raid. batteries.

it offers little performance benefit if any at all

all so you can have some outdated disk mirroring scheme

[devout ivy]

Many modern machines indeed hit disk bottleneck before CPU bottleneck if running as a dedicated storage server

even on quite puny CPUs

[long raft]

decryption, decompression, and reading the disk basically never saturate the cpu

you are very likely to be throttled by your 1gbps network if anything

[devout ivy]

start trying to push 10gb/s+ through tiered storage solutions with full encryption then you can start to make it sweat, but you arent gonna bust its balls

[long raft]

nah i cant.

[devout ivy]

and by it, I mean an intel atom XD

[long raft]

my server has an i3 i cant choke it with cpu stuff

[devout ivy]

But more seriously. Even the raspberry pi 4 chokes on I/O way before CPU

And realistically, that'll probably always be the case

[long raft]

my write speed is 400MB/s because the drives cant go faster, thats with encryption and compression

i promise the chokepoint will be the network if youre able to use ssd/nvme

[devout ivy]

Which I'd put under I/O

My own home network is only running on 1 gigabit switches

[long raft]

the PCI bus basically. yea i guess thats I/O

10gb is finally affordable, and also becoming necessary

[devout ivy]

And my desktop, thats on wifi at the moment

[long raft]

my wifi is > 1gbps, my internet provider will sell me 2gbps if i want, so ... itd all be wasted on a 1gb network

[devout ivy]

I have benched it at over 700mb/s throughput with my old card+access point setup

[long raft]

yea thats network throttle

well i dunno thats also similar to the speed of a magnetic disk, so...

[devout ivy]

That was just iperf testing

so true artificial load

[long raft]

100MB/s is like where i think of a 7200rpm drive on a 1gb network

[devout ivy]

Have since switched from ISP provided modem/router/AP combo to standalone modem+switch+router+access point

Access point is now wifi 6 2x2. But desktop is still on wifi 5 2x2

While I could swap its card out, I'd sooner just drill a hole in the ceiling and get it on ethernet

[long raft]

hm... how far is your run?

i thought 10gb was all about rj45, but ... no

[devout ivy]

10gb on RJ45 is possible with short runs

[long raft]

totally, 30m, but it gets hot

[devout ivy]

and tbf, I am talking almost literally just straight down through the floor

[long raft]

its hotter than my video card

nobody told me about the heat

[devout ivy]

so, about 1 storey's height worth of cable and maybe an extra meter at either end

[long raft]

the fiber cables are cool tho, and much cheaper if you go SFP+ ports

[devout ivy]

a 10gb run would be possible, but I don't currently have 10gb equipment

[long raft]

mikrotik sells a 4 port 10gb router/switch for like $110 but you have to buy the ports (or direct cable) additionally

[devout ivy]

and its unmanaged I think

[long raft]

nah its managed i thought

theyre pretty good

but again, hot

if i were wiring my place fresh, i would be thinking about twinax cable

[devout ivy]

Rental here

[long raft]

hmm

[devout ivy]

and not one I want to stay in for much longer

[long raft]

like if i was just drilling a hole and doing 1 cable, id be thinking twinax instead of cat7

[devout ivy]

Thing is in terms of drilling a hole and doing 1 cable. I already own a second managed switch

[long raft]

i just assumed 10gb was centered around cat7 and rj45 but found that wasnt as good, and was more expensive

oh

[devout ivy]

All my current gear is gigabit

and all of it is RJ45 only

[long raft]

also the network adapters for rj45 are kind of expensive

[devout ivy]

So right now we're talking sub 10m of RJ45 in a gigabit network on RJ45 equipment, its pretty plug and play

[long raft]

i just found sfp+ more versatile, i reused my existing cat5e in the walls with a rj45 port, but if i had like a 15 foot run id prefer to use twinax

and i just use twinax to the router because its next to the server, saved $100 and a lot of heat

[devout ivy]

Adding a proper NAS is when I'll want to go multigigabit, and I may just have to pull tricks off like giving the NAS 2 interfaces, 1 gigabit onto the switch and the other a 10gb static IP'ed point to point run

and then save obtaining a proper 10gb switch and doing proper runs for when I have my own place

[long raft]

well i dig that mikrotik router, very versatile, you can make it 1000gbps with $20 ports

[devout ivy]

interesting that it has dual redundant power input

[long raft]

i found network cards (that dont work on windows) for SFP+ i put on my linux server that were $25 for 4 of them

twinax cable is like 10-15 dollars

[devout ivy]

SFP cabling I'll probably just use the tooling at work

[long raft]

for 35 bucks i got my server connected to the router for 10gbps

[devout ivy]

so long as we don't take the piss, that generally flies

Currently got an ubiquiti edge router running OpenWRT, that again is fixed with just gigabit interfaces

though tbf, to get 10gb through to the NAS, not like my router needs to be gigabit

[long raft]

Oof i have one of those routers i hate it

Never knew there was an openwrt for it

[devout ivy]

All the ubiquiti gear tends to get openwrt rapidly

[long raft]

i hate that router

[devout ivy]

and tbf, theyve been "kind enough" to not pull a cisco and done nothing to block it

Albeit my modem is cisco

but I am on vdsl, so, not like I need gigabit throughput from the router

[long raft]

totally, just about NAS, and if its just multimedia 1gb is more than enough

my desktop doesnt have local storage

[devout ivy]

tbf if I roll my own NAS on linux, depending on the hardware in use and whether I put it upstairs or downstairs

I may even just use USB

RNDIS atop the gadget stack

Or, I've already had some experience with setting up mtp over USB on linux but mtp tends to be slow and the daemon for linux has quite a few bugs

[long raft]

why will the hardware matter

what os would you use otherwise?

[devout ivy]

USB device mode isnt typically available on your average intel box 😉

even many machines with a USB C port routed to the native type C interface of some newer intel chipsets tend to have BIOSes locking that interface into host only

though some machines have patches to switch it to device and/or DRP

On a pi compute module the usb interface can do device mode and therefore I could have the Nas direct usb plugged to my main machine for local access but that usb interface is only usb 2, so it would be pointless to do so

Realistically I'm probably best off just sticking a multi gigabit nic inboth desktop and Nas, just run a point to point link between Nas and desktop and the gigabit runs to the switch

[long raft]

or just go wireless

[pure sierra]

hdd update, i removed it from the mirror and put a blank partition on it and ran scandisk on it looking for bad sectors and now the health is 100% again. seems to have been a glitch

[long raft]

maybe bad controller or cable then

[pure sierra]

could have been a sleep/restore glitch

[long raft]

can it be resilvered?

[pure sierra]

?

[long raft]

re mirrored

[pure sierra]

dont see why not, but not sure if i want to bothher

[barren quarry]

@zealous sonnet the assets google drive has been made private apparently

(it says "no access")

pretty much the only thing i was looking for there was the color palette file

[long raft]

are you just looking for collie/warden colors?

[barren quarry]

yeah i guess

well i have some approximate values

[pure sierra]

its in the api readme but i believe its wrong

[barren quarry]

but i wasnt sure if theyre correct

[long raft]

its not wrong in the API, theyre just ... adjusted

srgb colorspace or something

[barren quarry]

oh indeed

they're in the api readme

[long raft]

or they claim to be but backwards

i have them as '#516C4B' and '#235683'

[lunar cobalt]

They're also in a pinned message in this channel

[hollow gyro]

The WorldExtent values, are they per hex, or the whole map?
I want to say the former, as I don't think these values changed when the 14 new hexes were added...?

WorldExtent.Minimum = (-109199.999997, -94499.99999580906968410989) WorldExtent.Maximum = (109199.999997, 94499.99999580906968410989)

Also I find the flags section utterly confusing, e.g. for MorCrossing all bases are 8 but the VP is 41 but this doesn't seem to match their README at all:

I know the numbers in the README are hex (1, 2, 4, 16, 32) but don't understand how they map to 8 and 41.

[pure sierra]

@hollow gyro, you need to learn bit masking and shifting, google it

you take the hex value for the flag given in the readme and use it as mask on the api flag value to get the result

basically the api flag value is actually bits, which each bit assigned to different flags, not decimal as you would see it normally

[hollow gyro]

So the value from the API is hexidicimal? (So 65 in dec)

[pure sierra]

if its 65 in dec, in binary its 0b1000001

each 1 or 0 relates to a flag

yes or no

the map flags section of the readme tells you whhat position relates to what

[hollow gyro]

I have a degree in ComSci but never did masking, but I know about the different base numbers.

If I wanted I can count to 31 on 1 hand 🙃

[pure sierra]

so 0x01 (isvictory) means the very far right digit

you can use masking to get that value computationally

[hollow gyro]

I see you've explained this before 🤦

[pure sierra]

so 0x01 & (and) 65 (0b1000001) = 1

so in this case is victory is true

if it was 0b1000000 is would be false

[hollow gyro]

So...basically turn the hex to decinal, that identifies which bit in the binary of the hex from the API means that it is that value.

So it's a cheeky way of returning multiple true/false values, but as a single hex number.

[pure sierra]

yes it called bit packing, and is a way to save data

[hollow gyro]

Alrighty I ubderstand that now, thanks fkr the explanation 👍

[hollow gyro]

What about WorldExtent?

They refer to a Hex, or the whole World Conquest Map?

[old ibex]

number* any number format works

[pure sierra]

im not really up on that, you dont really need it, i dont use it, but someone else can explain it

its for the whole world though i believe

im going to bed, bye

[hollow gyro]

Humm...109,000 yea seems world'ly

Thanks again 👍👍

Rest well

[hollow gyro]

Well yes.
I was just referring to their particular implementation; which is Hex.

[hollow gyro]

New question, is there a space for detailing these kinds of dicussions, e.g. a wiki that already exists?
I suspect not as it'd probably have been pinned.

[olive hamlet]

Is there any restrictions on what mods are allowed? I'm thinking of a mod that will make a 3rd party API call. It won't have an impact on gameplay, only visuals.

[languid harness]

Why you making an api call if its only visuals?

[pure sierra]

just here

[hollow gyro]

That kinda sucks

[pure sierra]

Its not a big demand topic and theres not a lot to it

[pure sierra]

im guessing anything that cold be regarded as cheating or bot like

[olive hamlet]

Player uploaded skins visible to everyone who has the mod. You upload your avatar, mod fetches skins based on steam ID from the API. Not sure how easy it is to implement on the client side yet, that's the next investigation task 😄

[languid harness]

You cannot pull IDs from the client afaik

Anything that isnt known by both factions via in game channels is not available in API

You'd have to use some sort of scraping system, which is illegal

[olive hamlet]

Sorry when I meant "from the API" I meant the custom one designed to power this mod

[olive hamlet]

Yeah that's what I was worried about, thanks for confirming though

[coral sundial]

Wat in tarnation

these arent even the same dang colors as they where in RGB

[lunar cobalt]

[pure sierra]

boogaloo

[hollow gyro]

What kind of coder takes a photo of their screen, instead of a screenshot?
There a security reason? 😀

[shadow rock]

drip

[hollow gyro]

heh?

[coral sundial]

@hollow gyro being lazy

Also i found the problem out

Its displaying hsv values as rgb

Which is interesting because you can clearly see the luminance mapped in green channel

Or saturation rather

[hollow gyro]

Fair 😄

[flint basin]

[fiery maple]

butt we but q be can u cc but box d cc ct xxx wvtctx cc cc,, AA as well wearing how BB cc h

[long raft]

what a day to be a college student. you can just ... download your textbooks from iran. could have got through school for half the price...

[lunar cobalt]

Nah they're still full price because all the homework is online locked behind single use keys or time based subscriptions @long raft

[long raft]

hah figures theyd find a way

so many perverse incentives behind text books

thats amazing, there was nothing like that when i was studying

[lunar cobalt]

just for this quarter I spent.. ~150 on online access codes

Just for homework

That's not even including the online textbooks, which are another like 90 or so

[long raft]

thats ... not too bad

i used to spend 300-400 per semester on books

with inflation is sounds like your costs are going down for textbooks

even without inflation, youre talking like $1000 a year for books, i spent at least 50% more than that, and they were used

and it was 1864 so a dollar was worth more

[lunar cobalt]

Spent 300 last quarter, but some of the books can be used across multiple classes which brought my price down this quarter

[oblique bane]

Well, I will say this. Eversus would be easy to take. There is one person building.

[pure sierra]

@oblique bane wrong channel ?

[oblique bane]

is it?

its a code.

[pure sierra]

how ?

[oblique bane]

I dont understand the code channel.

[pure sierra]

🙂

[oblique bane]

I thought we were supposed to talk about plans here. ._.

like radio codes and discussions. My apologies.

[pure sierra]

no, sorry, this is about programming and computer code

and anything tech

[oblique bane]

oh, then my apologies.

[pure sierra]

not sure if there is faction/team specific channels here ?

[lunar cobalt]

There are not.

[oblique bane]

was I lied to?

someone said this was all Warden.

[pure sierra]

no, but you can find tech support for foxholestats.com in #foxhole-discussion ;p

[lunar cobalt]

The main Warden-only discord can be found in #community-resources. As can the Colonial one.

[oblique bane]

Thank you ❤️

[pure sierra]

id never thought about it, but why dont that have faction chat here ?

i guess not secure

[lunar cobalt]

Because we have no way of guaranteeing that someone in one of those channels is supposed to be in that channel

[oblique bane]

True, like how I was lied to. Someone told me this was all Warden so I'm over here spilling the beans.

[pure sierra]

they might be implying mostly wardens in here, but its not the case

[oblique bane]

sorry again, you guys have a good night.

[cerulean urchin]

If we had faction channels here, it would be like the old days of global cross-faction chat lol

[pure sierra]

mud slinging

[chilly burrow]

why there is no ETag for
https://war-service-live.foxholeservices.com/api/worldconquest/war ?

[hollow gyro]

Because it only changes twice a war.
When it starts.
And when it ends.

So you need only request it twice-ish

[spark surge]

Hi guys, I'm a geologist and I'm interested in making a pseudo-geological map of the Foxhole game map. I'd like to get hold of height data to create a DEM (Digital Elevation Model). I want to get this data into some GIS software then populate the map with beaches, outcrops etc. that are in game. Does anyone know if it would be possible to extract height data from the WAR API? I'm not a coder so wouldn't know how to go about this. Also - I'd like to get hold of the game hex maps to import into GIS software as a basemap image. Any help or advice would be greatly appreciate. Feel free to DM me. Cheers!

[chilly burrow]

wasnt the whole point of ETags to reduce bandwidth, aka nobody knows when war is gonna end, so we have to keep calling that endpoint until something changes

[hollow gyro]

Well anyone can call any of the endpoints as much as they like.
However you should call them as often as is reasonable, e.g. every 10mins.
As for this /war one there's really no need for you to call it maybe more than once every hour; since it would only change on war-end.

[lunar cobalt]

There's some height maps that got posted at some point in this channel, but they're difficult to deal with because every map has a different baseline and scalar. So the heightmap for one map will have the same colors as another but at different elevations.

[pure sierra]

@spark surgestarts around here #code-talk message

[long raft]

i would have done the work to unify the height maps probably, but we are missing the 4 islands i think, so its not worth attempting

rust posted them, i havent seen them post the rest

[pure sierra]

we need this automated

it was so cool

[long raft]

i think we need the procedure to do it from rust, rather than the maps themselves

otherwise we'll never get updates

[pure sierra]

correct

needs to be scripted

[long raft]

manual is probably ok, i imagine it requires using unreal

i just dont want to try to find the mathematical solution for a one time thing, its gonna be a lot of work

[pure sierra]

no, there is standalone app for unpacking the pak file

[long raft]

i know but it wasnt just a simple asset stored inside

[pure sierra]

then just a matter of tilng them

[long raft]

he used UEviewer

said the island maps are not included because of a bug with ueviewer

[pure sierra]

let me try something else

[long raft]

"Extracted 'raw' tiles with vector info from game's files using UEViewer and then wrote a lil python script to 'stitch' them together..."

[pure sierra]

that python script would be handy

[long raft]

i bet its easy to rewrite

more concerned about the bug that prevents the islands from exporting

[pure sierra]

we pretty much do it already for warapi

i dont think that was the problem

[long raft]

#code-talk message

[pure sierra]

ok

[long raft]

#code-talk message

[pure sierra]

would /is FH UE4 ?

[long raft]

yes

[pure sierra]

im unpacking now

ok, so unpacking is one thing, but looks like still need ue asset viewer

[long raft]

yes

[lunar cobalt]

UModel exports those images FYI

[pure sierra]

thats what im using

[lunar cobalt]

ah cool

[pure sierra]

@long raft

[long raft]

he said raw files

[pure sierra]

that would be them no ?

[long raft]

nah, literal raw files

[pure sierra]

there are no raw files

[long raft]

these are TGA... theyre just textures

[pure sierra]

but they have the data

[long raft]

for sure? tga is easy to deal with

how do they tile together tho

[pure sierra]

let me convert them to jpg

[long raft]

i doubt you want to flatten the alpha out of them

[pure sierra]

ok png

[long raft]

i think they are a combination of light normals or bump maps and also height values

like maybe the alpha channel is the height

[pure sierra]

converting now, will take about 3 mins

only 1000 tga

[long raft]

you dont have paint.net?

[pure sierra]

gimp

[long raft]

it cant open tga?

[pure sierra]

yes, but easier to flick through in windows i thought

[long raft]

hmm

[pure sierra]

and get thumbnails

[long raft]

yea i dunno how they tile together

im afraid its something that isnt simple

[lunar cobalt]

UModel can also export as .png

[pure sierra]

oh neat

[long raft]

hopefully its just in column major order or something

[pure sierra]

how ? what is command

[long raft]

can tga hold a higher bit depth than 8?

nope.

itd be interesting if height turned out to be 2 of the components instead of 1, for more depth

[pure sierra]

likely its a standard ue thing you could google

[long raft]

but rust said he extracted raw files, and there is of course actual raw format that is often used for height maps and images

ive seen all these files before, i dont know how they tile together

youll find they arent even all the same size

[pure sierra]

so far all 64x64

[long raft]

theyre sort of clearly different data tho

hey rust can you enlighten us on this?

[long raft]

so you see anything that lines up in that?

oh i see 2 that line up

[pure sierra]

yep

[long raft]

which ones do you think have the height data

this appears to be lighting to me

here is the alpha channel

[pure sierra]

from that same bit above ?

[long raft]

yes

still kind of looks like lighting to me...

wut.

red

green

wow it hates blue

it wont let me post it. also blue looks like height lol

i cant tell tho

clyde thinks the blue channel is explicit content

yea i think we have the ... right ones i dont know which channel is the height tho

i think its the B or A

A makes the most sense but when i see it visually im not convinced

[pure sierra]

im still looking through fils

i found these files

[long raft]

might be best to start with deadlands

[long raft]

yea unprocessed

[pure sierra]

in a folder called textures/ui/hexmaps

[long raft]

yea

those are what im using for logiwaze now

i think some are ... outdated. i need to update fisherman's row

[pure sierra]

i think the file naming and output is generated by umodel/ueviewer, this is the output of list rather than just export

[long raft]

meaning?

[pure sierra]

maybe theres more data in the umap file than just tgas,

[long raft]

rust said he used those images, the purple hue ones

[pure sierra]

maybe some clues here https://github.com/cgcostume/pubg-maps

https://www.gildor.org/smf/index.php?topic=3229.15

[long raft]

#code-talk message

[pure sierra]

looks the same

[long raft]

is it 2 of the components that are height? maybe alpha and blue combined?

or likely just alpha

so my thinking is visually select all the maps like that, figure out how they line up, stich them, then extract the proper channel or channels for height

[spark surge]

@long raft , @lunar cobalt, @pure sierra - thanks for the responses guys. Looks like a complex problem. I know a few professional GIS guys that may have solutions to this, though I'm not 100% sure. I've read through from the older post by Rust and gathered that the maps are at different heights and scales for each hex, so stitching them together is an issue. I can throw out a message to the GIS guys for some help if you think it would be worth it?

[long raft]

sure. seems like a linear algebra problem to me

sample a pixel at each border, then solve the coefficients so they line up

i dont ... know if GIS guys would be able to help unless theyre good at math

step 1 is gonna be stitching the map, hayden. then it should be easy to extract the height from the combined version, and im pretty sure its in there if rust said it was

itd be funny if he was just using a normal vector component for height and didnt know better

[lunar cobalt]

I think he converted the normals into height maps somehow

[long raft]

from the light angle?

[lunar cobalt]

dunno

[long raft]

well there wouldnt be 4 channels in use in that image

[lunar cobalt]

I tried running one of the heightmaps to create a 3d model in blender and blender was very unhappy with me

[long raft]

they only need 3 for the normal, 2 in reality

did you extract a channel?

oh you used an actual provided height map

[lunar cobalt]

No

I used the image rust provided

[long raft]

they sure looked like height maps to me

[lunar cobalt]

Tried it on each channel separately, none gave me something that looked right

Then grabbed a grayscale to heightmap conversion plugin and it just got mad

[long raft]

no i think the heights are also scaled non linearly

[lunar cobalt]

Yeah agreed

[long raft]

that would sure make the math hard ... oof

i hope theyre all scaled the same way at least

[lunar cobalt]

My assumption is that this task is going to be as painful as possible, so they're probably not scaled the same

[long raft]

we will need surveyors

[pure sierra]

still got a problem with stuck G threads in apache

even after adding timeout to backend

whats strange is the pid listed dont seem to exist in linux

its as if apache thinks its finishing but in fact it already has

[spark surge]

What we need is an in game "survey" kit, to give height values - the only reason I'm asking in here (and I'm super happy I did because it's promising if you can extract height maps) is because of a silly idea to RP as military scientists/geologists analysing beaches and rock outcrops xD

[long raft]

any slow queries?

we all have different reasons for wanting height map data. artillery. roads. bunker construction

[pure sierra]

dont think so

[glad whale]

Well, I'll try to explain what I did

So, each tile has 4 channels

Alpha and blue channels contain normal data. Red and green is height data

To get normal is easy. You just grab those 2 vectors and fill blue channel with value of 1 (since blue channel in normal maps doesn't really matter)

But to get the height was a more pickle thing to do. Since I couldn't figure out what to do with the green channel

So I googled and found the pubg-maps GitHub that Hayden already posted here. https://github.com/cgcostume/pubg-maps

And I used the formula from there for the height calculating

Now the stitching process. Which is the main headache.

Last time I forgot to mention that I used Fmodel to get data for stitching (Since UEViewer can't do it)

So, using Fmodel I extracted data from .umap file as a .json file

And from that .json file I used the coordinates for each 64x64 tile

And by stitching them you get this

But then there's tiles that arent 64x64 scale for which you should make an exception and process them a bit differently

And then you will get this

But then there's another issue with the ocean area

Basically all those tiles are a baked texture of a landscape mesh. And in some regions the landscape mesh doesn't extend to the ocean areas

Thus some regions don't have tiles for the ocean area (In my script I filled such areas with the tiles of a value that I sampled from nearby tiles)

And as a result I finally got this

And speaking of the landscape meshes it's time for the problem with the island regions.

Basically most foxhole regions consist of just one landscape mesh and UEViewer exports them correctly

But there's a couple regions that consist of 2 landscape meshes: Westgate (1st Landscape Mesh is the Sanctuary area and 2nd Landscape Mesh is the rest of the region) and Shackled chasm (1st LM is the Simo's Run area and 2nd LM is everything else)

And if .umap file contains of more than one Landscape Mesh UEViewer exports tiles incorrectly

Basically some tiles of the LMs can have same name and when UEViewer exports them, they get overwritten

There's a simple way to get around this stupid bug but it works only in the case when there's only 2 LMs

And the reason I couldn't get the height maps for the island regions because they have way more than just 2 LMs. Fisherman's for example consists of 10 LMs. And Oarbreaker (if I remember correctly) consists of 16 (lol) LMs

Heres Fishermans LMs

[mossy pelican]

Any idea why they did that?

[glad whale]

It's easier to edit them I guess. Or maybe Matt is just a lazy person to make everything into one LM

But actually there's a dumb solution for this dumb problem I found (I can describe it later). It requires some manual labor but it's possible to get islands height maps (I did for fisherman's and it somewhat worked). The only downside is the water area (which doesn't really matter), gets a little scuffed in some places because of the same normalization problem

Here's the fisherman's region heightmap. I tried to "smooth" problematic areas in Photoshop but you still can see the obvious issue it has between the intersections of the LMs

I didn't do the other 3 island regions cause I don't have much enthusiasm left for it 😔

[long raft]

Thanks this is a ton of info. I had no idea it was so involved, but yea that answers some big questions!

[neat fossil]

From a production standpoint, there's literally no reason to make it one Landscape. That's just a massive, stupid waste of time.
Also, please refrain from the "Lazy Dev" rhetoric, it just makes you look like an asshole.

[long raft]

oh shit the devs read here

https://tenor.com/view/quick-hide-marothi-blanket-skeem-saam-gif-20190237

[glad whale]

Sorry Matt, I probably phrased it weird, I didn't mean to offend you (English isn't my native language). But yeah I agree, from the dev POV it's useless to wasting production time trying to combine them into one mesh.

[zinc osprey]

Dont think Matt got offended, its just every dev sees this "lazy devs" from people who know nothing of how complex everything is, and it eventually gets annoying. Tho i stopped caring.

[hollow gyro]

What people fail to see is a difference in knowledge.
Developers tend to know more than us, and the development process; what might seem lazy is actually genius because of something you just don't know.

The same observation could be made as a question and could have gotten an insightful answer.

[deft quartz]

There are many areas when it comes to Game Development. At the end of the day, its the players and the subjective interpretation of what is considered fun is what drives success/failures in a game. There's alot of topic on Game Design that can be interesting to see.

[hollow prawn]

Thank you!

[pure sierra]

this topic is dense

[neat fossil]

[quick acorn]

01000110 01010101 01000011 01001011 01011001 01001111 01010101 00100001

[long raft]

@lunar cobalt do you happen to have the last known good version of the reaching trail unprocessed map?

likely from september

[lunar cobalt]

Uhm I’ll get back to you in the morning

[long raft]

ok

[hollow gyro]

Well done 👍

[sinful steppe]

sus

[mossy pelican]

Thanks, i cant unsee it now

[lunar cobalt]

as far as I can tell my copies are all broken

[pure sierra]

hosting holding strong against new war ddos

[long raft]

nice

[pure sierra]

oh it would have gotten so salty

[long raft]

evidently looking sad is their only outlet

[hollow gyro]

War API - Wiki
Some additional writings, information, and neuances with the API are available here:
https://foxhole.wiki.gg/wiki/War_API

⬆️
Maybe someone could pin this?
I'll remove the embed soon

I assume <@&213133977681264640> can pin things; feel free to copy the content and then pin if preferred.

[pure sierra]

@hollow gyrothanks, ill see if i can add anything

[hollow gyro]

NP, it was enjoyable to write my learnings up 😄

[pure sierra]

link my git issue in "Map War Report" section about day of war

https://github.com/clapfoot/warapi/issues/81

[hollow gyro]

I didn't overly see it as a major problem myself, that's why I didn't include that one.

[pure sierra]

I have been getting a lot of complaints about it while i was using it, i recently switched to not use it, calculating it a different way.

maybe put it in the known issue bit at bottom than, as you have

[hollow gyro]

If anywhere that'd be the best place.
I just didn't see the need for knowing in-game days; it's an artificial construct that's not normally of a care to people; I'd assume.

[pure sierra]

as there are "improvement issues" and then theres the ones where the data is just wrong

[hollow gyro]

Bingo

[pure sierra]

I know, I actually just tried not showing it at all, as easier, but some users wanted it back so i hhad to correct it

I would also mention enlistments is only unique per region, we get plenty of confusion about that here

[hollow gyro]

Ah true

[pure sierra]

just 'found' another issue https://github.com/clapfoot/warapi/issues/99

"Combining this with World Extents will allow you to position structures onto a visual hex of a region if desired."

World extents is not used/needed at all for item placement

[hollow gyro]

Well no, but it helps with accurracy

Would it not?

[pure sierra]

how ?

i dont believe it has any relevance unless you want to represent a mapped distance in real world measuremnt ?

Another big issue people get stuck on you missed is the correlation between dynamic items and static labels, that you must find the closest one to know which belongs to which

' princpiples '

[hollow gyro]

Actually this is quite true, except that the extents gives you the dimensions so you can make the hexagon accuratly sized; without it you might make a one with equal height and width.

[hollow gyro]

That's true, that could go in the heading of 'Map Related Data' which precedes both 🤔

[pure sierra]

we just use the pixel sizes from the images as its more relevant

[hollow gyro]

I'll update the section to state it can help with accuracy and distances, but is not a blocker.

@pure sierra I haven't checked, do the static and dynamic instances of the same thing share the same x/y or are they slightly different?

[pure sierra]

they are different (i think)

[hollow gyro]

Cool 👍