#Core Haxe

1 messages · Page 4 of 1

velvet lodge
#

var data = Bytes.alloc(1024); - in a hard loop 🙈

#

(fixed now)

toxic stratus
#

I think I tripped that at some point lol

velvet lodge
#

yeah, sorted now

#

i think i might have fixed the recusion too... the problem is that the socket would constantly call handleData if there is always data coming through, so if the server was constantly sending data then, effectively, the socket would sit there in a while(true) pumping data to the buffer constantly and calling handleData with no respite

#

the fix would be to iteratively build up the buffer in the process() call, which seems to work... just need to see if it breaks other things

#

alright, if you pull latest can you try again at some point, i think it should be fixed...

#

ive also changed the way it uses the main loop, so by default, should work nicer too (you can activate the threaded behaviour with -D websockets_threaded

#

hard to tell for sure, but feels faster too

toxic stratus
#

want me to pr those?

#

or i guess it's probably better to make a change to the existing pr

velvet lodge
#

yeah, if you could, ill take a look, but they need unit tests before incorporate the changes

#

more interested to see if the initial recursion issue is sorted tbh

toxic stratus
#

will check now

toxic stratus
#

in my app this is the deepest the socket stuff will go now

#

will try threaded as well

#

something is wrong with threaded mode

#

app basically goes to a standstill

#

i wonder if this is my app or the sockets lib

#

will go to the smaller test case i made the other day

velvet lodge
#

does it work though in the first image?

toxic stratus
#

setting the threaded flag it fails

velvet lodge
#

ie, the first image?

toxic stratus
#

yea

velvet lodge
#

perfect

#

might have broken something with threaded, not sure, but i think the main loop method (ie, not threaded) is a better way to use it anyway

toxic stratus
#

compiling sleep(0.1) atm to see if that helps

velvet lodge
#

its not relevant anymore in the unthreaded version

toxic stratus
#

i mean for the threaded version

velvet lodge
#

right

toxic stratus
#

it does seem to help

#

got a null reference i have to sort though

velvet lodge
#

so, anyway, the initial recursion issue is sorted 👍

#

(and that OOM issue too... )

toxic stratus
#

correct!

#

the detour was much appreciated thanks! 😄

velvet lodge
#

np - defo worth fixing - and fyi - the same issue applies to hxWebSockets

#

probably hard to tell, since websockets are pretty nippy, but does it feel any faster?

toxic stratus
#

it's hard to tell because my app is a bit sluggish in general on cpp

velvet lodge
#

right

toxic stratus
#

my adventure will probably lead me back to my haxeui backend

#

I never really used profilers much because they never seemed that useful

#

my experience was only really with chrome's and maybe it's a skill issue

#

but here things seem so obviously wrong

#

but i guess, on js you can get away with a lot of bad

velvet lodge
#

true enough

toxic stratus
#

was threading on or off before?

velvet lodge
#

on

toxic stratus
#

ah okay cool

velvet lodge
#

the default is the main loop now, which i think is better - i kept the threading there though anyway (behind a define) just in case

#

the hard loop is there also btw (-D hard_loop) but i think that is going away very soon 🙂

# 
#if hard_loop
while (true) {
    read = _socket.input.readBytes(data, 0, data.length);
    if (read <= 0){
        break;
    }
    //Log.debug("Bytes read: " + read, id);
    _buffer.writeBytes(data.sub(0, read));
}
#else
read = _socket.input.readBytes(data, 0, data.length);
if (read > 0) {
    _buffer.writeBytes(data.sub(0, read));
}
#end
toxic stratus
#

main loop is slower than threading for me but i guess that's expected?

velvet lodge
#

possibly if you are doing other things (which i guess you must be)... you might be able to bump the priority

#

depends what is more important: comms or render

toxic stratus
#

they're equally important 😂

#

the render is based off the incoming

velvet lodge
#

i doubt thats accurate

#

my guess is data is more important (ie, comms)

#

nothing to render without data at the end of the day

#

anyway, threading should still be a viable option, havent checked why its not working for you, its defo possible i borked something (i wasnt trying to work on this at all this weeked)

toxic stratus
#

I've never heard of priority in this context, how is that a thing?

velvet lodge
toxic stratus
#

the most important thing is, everything runs now on the main target I want to use

toxic stratus
velvet lodge
#

yeah, ill check out threading again at some point, but this currently is a lib i simply dont need - that said, it was nice to get the recursion and and OOM issues out the way for when i do need it

velvet lodge
#

i dont know how it uses the MainLoop, might not be via the add(event) at all

toxic stratus
#

so interesting

#

I'll have to take a look at some point

velvet lodge
#

i would assume it uses its own event loop (via SDL or something) and simply "ticks" the haxe event loop

toxic stratus
#

i just caught this in tracy

velvet lodge
#

if that is the case, then yeah, threads would presumably be much faster

velvet lodge
toxic stratus
#

the stack continues that deep

#

but it might be my fault

#

not completely clear

#

i don't know how these stacks get generated and what constititutes a "new stack"

velvet lodge
#

thats not that deep... it just means it can process more data in that window (timewise)

#

the other was actually recursive

#

thats actually a "nice" stack, because it means it didnt have to wait for another "tick" to process more data

toxic stratus
#

ah, yeah

velvet lodge
#

even though that looks nicer, actually, thats a "less effecient" frame (from the PoV of websockets)

#

it had more data, but had to wait to process it

#

you have "one thing" filling up a chunked buffer of bytes (the socket) and then "another thing" processing those bytes... if they can all happen at the "exact" (not possible) time, that is ideal

toxic stratus
#

super interesting

#

you get a lot more out of the profiler than i do 😄

velvet lodge
toxic stratus
#

yeah, i think hard loop wouldn't work but i can give it a try and see

#

the market is always active

velvet lodge
#

"hard loop" was the original, the one where you got that huge stack

toxic stratus
#

ahh

velvet lodge
#

i just never tested with a server just constantly pushing data... there was always a "request", "response" kinda thing going on

#

so defo a bug, and pretty good one

toxic stratus
#

that makes sense

#

i guess it would still benefit your use case as well

#

my use case just makes it very clear that there's an issue

velvet lodge
#

100%, core-haxe/ws has defo benefitted from your findings, for sure

#

(not to mention to the alloc(1024) in a hard loop 😮 )

#

that was a crazy one btw

#

if you have a client socket (in non-js) and it doesnt get any data, it will sit there create 1kb of data every... god knows... nano second

# 
while(true) {
   Bytes.alloc(1024);
}

basically... ... ... crazy

toxic stratus
#

lol

#

i never investigated those issues

#

typically i'm recompiling my app before i triggered it so i never thought it was an on going issue

uncut phoenix
#

@velvet lodge Am I right to assume that if find() doesn't find anything, it will reject?

velvet lodge
#

nope, it will return a null entity

uncut phoenix
#

I am confused on why onRejected wants a function that returns my entity

#

ah

velvet lodge
# 
MyEntity.find(101).then(entity -> {
    trace(entity == null);
}, error -> {
    trace("error", error);
});
#

because it may be perfectly valid to not find an entity

uncut phoenix
#

yeah, makes sense

#

thanks

#

this is neat

velvet lodge
#

yeah, its pretty cool - and kinda mental powerful when you get "into" it

uncut phoenix
#

ah yes, uknonwn
sqlite/impl/cpp/SqliteDatabase.hx:149: UKNONWN:,TClass(entities.User)

#

hm does the sqlite backend not support relations?

velvet lodge
#

defo should

#

is entities.User an entity?

uncut phoenix
#

yeah

#

(probably a bad idea to make the package they reside in also called entities)

velvet lodge
#

doubt thats it

#

what do your entities look like? Its looks like its coming from core-db, not entities

#

sqlite is defo less tested than mysql, but should be fine

uncut phoenix
# 
class Session implements IEntity {
    @:cascade(deletions) public var user: entities.User;
    @:size(36) public var accessToken: String;

        // ... static functions
}

@:exposeId
class User implements IEntity {
    @:size(254) public var username: String;
    @:size(254) public var hash: String;
    @:size(24) public var salt: String;
        // ... static functions
}
#

nothing too complex

velvet lodge
#

what if you dont use the fully qualified path?

@:cascade(deletions) public var user:User;
#

?

uncut phoenix
#

nope same

velvet lodge
#

very odd, its a super weird error too

#

what haxe target, and can you paste some test code using these entities so i can test my end?

uncut phoenix
#

I am using hxcpp and yeah give me a bit

#

by code you mean just a simple main class that connects to a db and adds a users and attaches it to a session?

velvet lodge
#

yeah, basically

#

just something i can run this end to see if i can repro

# 
class Main {
    static function main() {
        trace("running");

        var db = DatabaseFactory.instance.createDatabase(DatabaseFactory.SQLITE, {
            filename: "test.db"
        });
        EntityManager.instance.database = db;
        var session = new Session();
        var user = new User();
        user.username = "tim";
        session.accessToken = "bob";
        session.user = user;
        session.add().then(_ -> {

        }, error -> {
            trace("error", error);
        });
    }
}

class Session implements IEntity {
    @:cascade(deletions) public var user: User;
    @:size(36) public var accessToken: String;
}

@:exposeId
class User implements IEntity {
    @:size(254) public var username: String;
    @:size(254) public var hash: String;
    @:size(24) public var salt: String;
}
#

this seemed fine

#

also, totally unrelated, but do you really want @:cascade(deletions) public var user: User;?

#

this means when you delete the session, it will delete the user too - might be totally what you want, but feels like you might be accidently nuking your user data 😮

uncut phoenix
velvet lodge
#

nope, it will cascade deletions through

uncut phoenix
#

anyways trying your code now

velvet lodge
#

using entities2, right?

uncut phoenix
#

correct

velvet lodge
#

latest git version

uncut phoenix
#

correct

#

it works hmm

#

i even tried splitting it up into multiple files and it works

velvet lodge
#

i wonder if it is the package... lemme try

#

also fine

uncut phoenix
#

hm

velvet lodge
#

what does your code look like? Is it paste-able?

uncut phoenix
#

i have it all on github here

velvet lodge
#

haxe_modules/entities

#

defo sure thats entities2?

uncut phoenix
#

yes

velvet lodge
#

cool

uncut phoenix
#

hm i think it happens when Session.findByCredentials gets ran

velvet lodge
#

so, i dont know if its the actual error, but there are a bunch of times that you are using .add and .update and arent wating for the promise to return

#

but i dont actually think thats the issue tbh

uncut phoenix
#

oh ill fix that

velvet lodge
#

hang on, i was using entities1 😄

uncut phoenix
#

i tested with entities2 and it worked fine for me

velvet lodge
# 
./src/Main.cpp(69): error C2860: 'void' cannot be used as a function parameter except for '(void)'
#

:/

#

ok, also worked with entities2

uncut phoenix
#

try doing a session find with the user as a query?

velvet lodge
#

OK, i can see what it is 🙂

#

yeah

# 
sqlite/impl/cpp/SqliteDatabase.hx:149: UKNONWN:,TClass(entities.User)
# 
    public static function findByCredentials(user: entities.User, accessToken: String) {
        return find(Query.query($user == user && $accessToken = accessToken));
    }
uncut phoenix
#

oh the single equals lmao

velvet lodge
#

the query needs to be the userId, so:

# 
    public static function findByCredentials(user: entities.User, accessToken: String) {
        return find(Query.query($user == user.userId && $accessToken = accessToken));
    }
uncut phoenix
#

aaah

#

let me try now

#

yup, works. thanks so much!!

#

i guess i didn't think about that

velvet lodge
#

np - its kinda gotcha because it indicates having knowledge of the table structure (which ideally you never touch in entities), but im not sure there is a good way around it

#

i mean, is it safe to assume if you mentioned another entity in the function param you always mean the primary key? Im not sure it is

uncut phoenix
#

Could I write a custom equals function?

velvet lodge
#

... .. maybe it is 🤔

velvet lodge
uncut phoenix
#

If entities supported it, maybe in users I could do something like

public function equals(u: User): Bool {
  return userId == u?.userId;
}
#

to override the default functionality

velvet lodge
#

thats not supported no in query-dsl, not sure how it would help tbh either

#

im kinda thinking thats its safe, for entities to find this function:

    public static function findByCredentials(user: entities.User, accessToken: String) {
        return find(Query.query($user == user&& $accessToken = accessToken));
    }

check the params for entities, and that its used in a query(), and map the primary key in the query (ie, rewrite it at the macro level)... but ill need to think about it because it could be a foot gun

uncut phoenix
#

is there a difference between == and =

velvet lodge
#

nope

velvet lodge
uncut phoenix
#

why is the id not exposed by default?

velvet lodge
#

because it goes "against" the idea of entities... the idea that you just dont think about things like that anymore, just your haxe classes, ofc, sometimes you do need the id (which is why that meta exists), but for example, in the WHOLE of glint, thats the first time ive accessed the primary key directly

uncut phoenix
#

hm interesting

velvet lodge
#

and by "against" i dont mean "you cant or you shouldnt", just that, the spirit (ideally) is you dont think about "whats the id of this"

#

and actually, i think i could fix my usage with a "sync" function

#

the problem here is the userToUpdate could just have and primaryKey, and lets say, username

#

actually, you know what, im talking shit

#

i think i can just call "userToUpdate.update()", it will only update things that exist

#

i thought it would nuke things, but why would it... .. 🦆

#

yeah, works fine 🙂

uncut phoenix
velvet lodge
uncut phoenix
#

Out of curiosity, how many people are working on Glint?

#

It kinda confuses me, I am on the website and there is like 10,000 people in the discord server but only like 70 people online

velvet lodge
#

2 devs, whole marketing team (prolly 5 people), few (new) testers - i think 3, then i woud say, dunno, 3-4 other people (various hats)

#

we did have some "data engineers", but they are a joke, like total joke

uncut phoenix
#

lol

#

feels like they could move some of the energy from marketing to development, but I guess they have good devs at least

velvet lodge
#

its getting there (this is just a dashboard i knocked up earlier)

image.png image.png
uncut phoenix
#

It looks really pretty

velvet lodge
#

yeah, its come along way in a semi short amount of time

#

the stack is mental nice to work with - so easy to add new bits (client / server / job consumers, its all super modular and super easy to plug bits in)

uncut phoenix
#

all haxeui+core right?

velvet lodge
#

yeah, and a bunch of other libs that may (probably will) go into core-haxe / haxeui (as non core things) at some point

uncut phoenix
#

👀

#

i think i got everything ported over from pure db-core -> entities now

#

outside of the find thing, it was really smooth

velvet lodge
velvet lodge
# uncut phoenix i guess i didn't think about that

quick fyi, with (alot of) help from the #macros 's guys this now works, you can still use the user.userId, but if you just use the entity, it will auto replace the primary key in the query - i just didnt like that your inital "hunch" was, imo, the right one, and that entities2 forced you down the "you must know about the table structure route"

toxic stratus
#

I really wanna use entities but I'm stuck on this performance/optimisation path atm :/

uncut phoenix
#

Feels a lot simpler. With db-core I had this giant class where I handled all of the db stuff. I wasn’t really sure how I was going to split it up. Entities forced me (in a good way) to split things up sanely.

velvet lodge
#

nice - defo the goal - the idea is that you basically just "stop" thinking about the db part - which for the most part we dont care about

uncut phoenix
#

It doesn’t feel databasey at all, even compared to other ORMs I have used.

uncut phoenix
#

I am guessing the latter?

velvet lodge
velvet lodge
uncut phoenix
uncut phoenix
#

what is the best way to break out of a promise .then() chain if something goes wrong?

glacial relic
#

isn't there normally a .catch() you can add to the chain to handle errors ? Not my area but that's what I recall from a a brief bout of js at one point

uncut phoenix
#

there is a second parameter in .then for errors

#

onRejected

glacial relic
#

Oh ok

velvet lodge
spiral lava
#

Is there any minimal entities2 usage sample? I'm a little confused how to sync my entities with the databases, i. e. something like this in sqlalchemy:

Base.metadata.create_all(engine)
velvet lodge
#

You dont sync, thats what entities2 does:

EntitiesManager.instance.db = ...
...
...
var myEntity = new MyEntity();
myEntity.someField = 1;
myEntity.add().then(_ -> {
   ...
}, error -> {
   ...
});
spiral lava
#

ah, ok, missed the EntitiesManager.instance.db = ... part

velvet lodge
#

if you wanted to add a bunch of entities at once you can do something like:

var promises = [];
for (i in 0...100) {
  var myEntity = new MyEntity();
  myEntity.someField = i;
  promises.push(myEntity.add.bind());
}
PromiseUtils.runSequentually(promises).then(_ -> {
   ...
}, error -> {
   ...
});
spiral lava
#

Yep, but I mean, creating a table if it doesn't exist (when I add a new entity class)

velvet lodge
#

table creation is handled automatically (as is table modification), eventually, there will be a system for upgrading the tables (to avoid data loss)

spiral lava
spiral lava
velvet lodge
#

also, table creation is lazy, its only when you use the entity, however, you can init all entities at once with EntitiesManager.instance.initAll (or something like that, dont remember the call)

spiral lava
#

nice, that's what I was looking for, many thanks

#

EntityManager.instance.initAllEntities()

velvet lodge
#

extract from a server i use:

public static function initEntities(config:ConfigNode, path:String = "/database/connection"):Promise<Bool> {
    return new Promise((resolve, reject) -> {
        findEntitiesDatabaseConnection(config, path).then(database -> {
            EntityManager.instance.database = database;
            return EntityManager.instance.initAllEntities();
        }).then(_ -> {
            resolve(true);
        }, error -> {
            reject(error);
        });
    });
}
#

(initAllEntities)

spiral lava
#

thanks

#

As I can see, normal enums are not supported, only enum abstracts, right?

velvet lodge
#

yeah, looking to support enums at some point, just not 100% what the table structure would look like... cant just be an index because that could change... its possible i think to use a string or something (ie, the enum item name) but it feels like a hack

#

... ... would make querying simpler though

spiral lava
#

if it's of any help, sqlalchemy uses names by default, but it's possible to set a certain flag to use values instead (translating into haxe, that means instance names by default and instance indexes as an opt-in)

#

that makes much sense, considering index may be accidentally or intentionally changed much more often than name

velvet lodge
#

yeah, ill have a think - probably name is the quickest path - but not something that is a priority for me currently

velvet lodge
#

select * from X where enum_field='enum_value'

spiral lava
#

yep

spiral lava
#

And, in turn, for enums with parameters I would expect auxiliary tables to be created. Something along the lines of:

enum SomeEnum
{
    NoArgs;
    OneArg(a:Int);
    TwoArgs(b:String, c:Bool);
}

class SomeEntity implements IEntity
{
    public var field1:String;
    public var field2:SomeEnum;
}

->

CREATE TABLE some_entity (
    some_entityId INT NOT NULL AUTOINCREMENT
    field1 TEXT NOT NULL,
    field2 ENUM('NoArgs', 'OneArg', 'TwoArgs') NOT NULL
)
PRIMARY KEY (some_entityId);

CREATE TABLE some_entity_field2_one_arg (
    some_entityId INT NOT NULL AUTOINCREMENT,
    a INT NOT NULL
)
PRIMARY KEY (some_entityId)
FOREIGN KEY (some_entityId) REFERENCES some_entity (some_entityId);

CREATE TABLE some_entity_field2_two_args (
    some_entityId INT NOT NULL AUTOINCREMENT,
    b TEXT NOT NULL,
    c BOOL NOT NULL,
)
PRIMARY KEY (some_entityId)
FOREIGN KEY (some_entityId) REFERENCES some_entity (some_entityId);
#

implies joins when filtering by the enum field value:

...someEntity.field2 == SomeEnum.TwoArgs('abc', true)...

...
FROM some_entity AS t1
INNER JOIN some_entity_field2_two_args AS t2
ON t1.some_entityId == t2.some_entityId
WHERE t1.field2 == 'TwoArgs'
AND t2.b == 'abc'
AND t2.c
...
velvet lodge
#

nope - not gonna happen 🙂

spiral lava
#

why tho 4illFeelsRainMan

velvet lodge
#

because its adding alot of complexity for a relatively small feature - and its one that i dont even (personally) need atm

spiral lava
#

In entities2, can I specify the size for an integer column (tinyint/smallint/int/bigint)? Can I add UNSIGNED restriction?
EntityBuilder seems to only respect the size for String fields, but not Int. I see no mention of unsigned throughout the whole source code either

#

Of course, that would be backend-dependent, but still, it's a useful feature to have

velvet lodge
#

so yeah, size will (at some point) be used for things other that string - its supposed to (eventually) be sort of generic meta - as for unsigned, thats not a thing, no, it would have to be implemented in db-core first and then exposed to entities

spiral lava
#

Thanks! What about providing the default values for the columns? Is it possible?

column1 INTEGER DEFAULT 0,
column2 TIMESTAMP DEFAULT CURRENT_TIMESTAMP()
velvet lodge
#

nope, also not possible currently, but it would be very useful

spiral lava
#

agreed

velvet lodge
#

ill take a look at it at some point, too busy creating incredibly elegant user defined themes for my application

image.png
#

😄

toxic stratus
#

😂

knotty vortex
toxic stratus
#

is there anything special I need to run db-core/mysql on linux?

#

I'm trying to compile and I'm getting

/usr/bin/ld: cannot find -lmysqlclient: No such file or directory
collect2: error: ld returned 1 exit status
Failed to compile C++

#

but the library definitely exists

uncut phoenix
#

Make sure it is the devel version

toxic stratus
#

oh there's something i need for linux?

uncut phoenix
#

Oh. Maybe not. Looks like it is vendored

toxic stratus
#

the CI runs it all on linux fine i suspect

#

i can't imagine they're running on windowu

#

yeah, it's running on ubuntu

#

what does this "super hack" mean

#

the thing in brackets seems interesting

#

could that be the missing link i require?

#

i installed mysql-devel

#

Compile C++
Using libmysqlclient from system
Link: Main-debug
/usr/bin/ld: cannot find -lmysqlclient: No such file or directory
collect2: error: ld returned 1 exit status
Failed to compile C++

#

now I get that

#

a Google suggests that Ubuntu and fedora have different names, would that cause an issue with externs? 🤔

uncut phoenix
#

should come with mysql-devel

#

not sure why it isn't being found by libmysqlclient tho

toxic stratus
#

it's defo here

#

oh

#

maybe it's a atomic issue

#

it isn't available in the "normal" mode

#

but it's available in the container

uncut phoenix
#

i am trying too

#

on non-atomic

toxic stratus
#

in the meantime i'm gonna install it in "normal mode"

#

didn't make a difference

uncut phoenix
#

@toxic stratus not the most elegant, but sudo ln -sf /usr/lib64/mysql/* /usr/lib64/ fixes it

#

not sure why the linker doesnt pick up stuff in /usr/lib64/mysql

toxic stratus
toxic stratus
uncut phoenix
#

i use sqlite, i was just messing with it now

toxic stratus
#

it finishes but it finishes with an error code somewhere?

uncut phoenix
#

umm that seems like a ceramic thing

#

compiles no error here

#

but i am not using it in a ceramic project

toxic stratus
#

app runs and i get some traces but it stops responding

#

might be a me issue here

#

db doesn't connect though

uncut phoenix
#

hard to say without trying it myself

toxic stratus
#

might be a case of my ip has changed so server is blocking the request

#

it just fails differently on linux

#

ahah, ipv6 is active

#

need to find out how to disable that x)

#

yep, that was exactly it

#

I'm in. Thanks @uncut phoenix

uncut phoenix
#

np

velvet lodge
#

so... ... all good?

uncut phoenix
#

Yeah. Just Fedora weirdness I guess.

toxic stratus
#

I'll post a pr for that in the readme at some point

frosty locust
#

The libmysqlclient issue should be solved by modifying build.xml (to add a -L path) rather than by creating a symlink

uncut phoenix
#

I tried messing with ldconfig or whatever but not sure that is applicable

frosty locust
#

You could allow setting a -D MYSQL_LIBPATH define that is added as a linker path

#

Also, if it is using the system library, it should also be using the system headers

velvet lodge
#

i cant remember exact the call now, but there was a linux call you could make (similar to wxConfig) that gave you all the compiler settings

#

by call i mean process

#

in hxWidgets we make the call in the macro, parse the results and dynamically build the hxcpp xml

# 
mysql_config --cflags --libs

?

toxic stratus
frosty locust
#

That would be the right thing to use yeah

velvet lodge
#

yeah, then use the results of that (in a macro) to dynamically create the build xml - hxwidgets does it, and it seems pretty flawless

uncut phoenix
#

Been trying to do the same thing with linc_dialogs but it wasn’t working for me

#

Might need to get you to look at it if you ever have time.

velvet lodge
toxic stratus
#

The process doesn't seem that bad

#

it looks like you're just constructing the xml in the macro

#

and inputting some command responses

velvet lodge
#

exactly that yeah, based on the results of a process call

toxic stratus
#

so find a working configuration, find the appropriate system calls/flags to get the appropriate info

velvet lodge
#

and i think that for the mysql one its even more simple since i dont think you have to take "other" things into account

toxic stratus
#

i doubt you'd even need much macro stuff here

velvet lodge
#

well, it has to happen in the macro to build the xml meta before any compilation starts

toxic stratus
#

it's a pretty neat way to handle it

velvet lodge
#

i think ibilion came up with it (iirc)

uncut phoenix
velvet lodge
#

any error? code?

uncut phoenix
#

the weird thing is if i run the command on line 92 myself and recompile it works fine

hazy mulch
#

is there a way to use http without async?

#

i know normally thats bad but im trying to make a script to mass delete spam accounts on my gitlab instance (over a thousand of them), and i just want it to do one at a time so it doesnt kill the instance

toxic stratus
#

don't spam them all at once

#

one thing you can do is, don't send another request until you recieve the response from the last one as an example

hazy mulch
#

the requests just never seem to complete what

toxic stratus
hazy mulch
# 
trace('deleting ${user.username} (${user.id})');
    client.delete('https://git.team-prism.rocks/api/v4/users/${user.id}', null, ['hard_delete' => ''], ['PRIVATE-TOKEN' => 'glpat-'])
        .then(response -> {
            trace('Deleted ${user.username} (${user.id})');
            deleteNextUser();
        }, error -> {
            trace(error);
            deleteNextUser();
        });

this is what im doing right now, and it traces the "deleting" line, but then just stops. ive let it sit for like 2 minutes at least

nh4iilx.png
#

i would expect it to at least error out but it just isnt

#

maybe i dont know how to read gitlab logs but it doesnt appear to show up there at all

toxic stratus
#

is the app just exiting after it runs that command?

hazy mulch
#

but yeah, there is a loop going

toxic stratus
#

awkward to help because i don't have access to the api stuff

hazy mulch
#

also fwiw i have latest versions of everything, since i downloaded it today

toxic stratus
#

I see some odd things in the logic

#

but want to try it first

toxic stratus
# 
    var timer = new Timer(100);
    timer.run = () -> {}
#

you're blocking the application from running

hazy mulch
#

thats unintuitive but it works

hazy mulch
#

i keep getting Thread 28 killed on uncaught exception Unix.Unix_error(3, "recv", "")

#

which causes it to freeze again

#

but i can just kill it and continue where it left off

velvet lodge
#

hxcpp / hl i assume?

#

ok, i think i can see what NB meant, that while () loop isnt a great idea, although you do have a sleep in there, but still, not the best way, two secs

# 
class Main {
    static var client = new HttpClient();

    static function main() {
        trace("started");

        deleteAllUsers().then(results -> {
            trace("all operations complete", results);
        }, error -> {
            trace("error", "error");
        });
    }

    static function deleteAllUsers():Promise<Array<Int>> {
        return new Promise((resolve, reject) -> {
            var url:String = "http://google.com"; // just for illustration purposes
            var urls:Array<String> = [];
            var results:Array<Int> = [];
            for (i in 0...10) {
                urls.push(url + "?q=" + i);
            }

            deleteNextUser(urls, results, resolve, reject);
        });
    }

    static function deleteNextUser(urls:Array<String>, results:Array<Int>, resolve:Array<Int>->Void, reject:Any->Void):Void {
        if (urls.length == 0) {
            resolve(results);
        } else {
            var url = urls.shift();
            trace("making request", url);
            client.get(url).then(_ -> {
                results.push(Std.random(0xff));
                deleteNextUser(urls, results, resolve, reject);
            }, error -> {
                trace("error", error);
                results.push(-Std.random(0xff));
                deleteNextUser(urls, results, resolve, reject);
            });
        }
    }
}

( @hazy mulch )

hazy mulch
velvet lodge
#

right, should be the same anyway

#

yeah, its fine:

# 
src/Main.hx:10: started
src/Main.hx:48: making request,http://google.com?q=0
src/Main.hx:48: making request,http://google.com?q=1
src/Main.hx:48: making request,http://google.com?q=2
src/Main.hx:48: making request,http://google.com?q=3
src/Main.hx:48: making request,http://google.com?q=4
src/Main.hx:48: making request,http://google.com?q=5
src/Main.hx:48: making request,http://google.com?q=6
src/Main.hx:48: making request,http://google.com?q=7
src/Main.hx:48: making request,http://google.com?q=8
src/Main.hx:48: making request,http://google.com?q=9
src/Main.hx:24: all operations complete,[101,129,246,243,45,75,178,214,83,131]
wide solstice
#

out of interest because it was suggested in haxeui channel

#

why is websockets far away from being complete?

#

is it because of the incomplete server?

#

another question: Is the client usable with javascript target, because I'm targetting mobile and html5 at the moment

toxic stratus
#

I think it's mostly a question of "Ian time" than anything else

velvet lodge
#

pretty much yeah, unlike hxWebSockets, the core-haxe version will all be unit tested, etc... so everything takes more time because as im migrating it across, im writing tests, making changes, etc - so its not quick. Add to that, that im totally swamped recently (and the fact i dont actually need it... ... ... yet)

velvet lodge
velvet lodge
toxic stratus
#

yeah, it's mostly been fine but I've been on a few side quests the last few weeks

velvet lodge
#

yeah, its very much a "use it at your own risk, i dont have bandwidth to support it currently" kinda thing

toxic stratus
#

completely fine for dev work

#

I'd say node target is completely fine tho

velvet lodge
#

its probably not a million miles away to be fair, all the client side unit tests are running for all targets just fine

#

filename is wrong, just realised that 🙂

#

actually, not they are alright... i just have unit tests for the test server too just to make sure it worked

wide solstice
#

What needs to be done for html5/js target? I see that there is a test-js.hxml but I guess it's not yet prepared correctly. Would love to contribute a bit further 🙂

toxic stratus
#

both server(nodejs) and client(js & nodejs) work iirc

wide solstice
#

okay, then I can experiment a bit with that client to get rid of haxeui hack executing in main loop 😄

wide solstice
#

sad, that haxelib.json doesn't support dev dependencies. Have to install promises library since it's necessary for the tests 🙂

toxic stratus
#

it does I just think there was some issue it caused

toxic stratus
#

Are there any dependencies one requires for mysql on windows, seem to get stuck at "trying to connect". The details work on electron and in heidi. Maybe i'm missing a windows thing?

#

Getting no errors or anything

#

bit odd

velvet lodge
#

you need libmysql, but that should be part of it

toxic stratus
velvet lodge
#

what does your connection code look like?

toxic stratus
# 
    function connect() {
        trace('attempting to connect to db');
        db.connect().then(function(state) {
            if (state.data) {
                this.connected = true;
                this.connected_time = Date.now().getTime();
                trace('Database connected');
            } else {
                this.connected = false;
                this.connected_time -= (Duration.fromString('5hrs') : Float);
                trace('Database not connected');
            }
        }, (err) -> trace(err));
    }
#

works completely fine on nodejs/electron

velvet lodge
#

sorry, i mean the details

#

ie, the creation of db

toxic stratus
# 
        db = DatabaseFactory.instance.createDatabase(DatabaseFactory.MYSQL, {
            database: database,
            host: host,
            user: user,
            pass: pass
        });

        db.setProperty('autoReconnect', true);
        db.setProperty('autoReconnectInterval', 5000);
        db.setProperty('replayQueriesOnReconnection', true);
velvet lodge
#

seems fine 🤔

toxic stratus
#

i'm sure it did work at some point, but for some reason nothing now

#

this is a new install of windows

#

but I just get stuck at attempting to connect to db system/DatabaseSystem:63 and nothing happens after that

#

no errors, no missing dlls no failed to connect messages

velvet lodge
#

i would try and run the unit tests in libmysqlclient, make sure they work

velvet lodge
toxic stratus
#

it is, heidi connects to 3306 successfully

velvet lodge
#

whats heidi?

#

and yeah, you said its fine for nodejs

#

so yeah, best advice is to to see what the unit tests say from the lowest possible part of the stack (libmysqlclient)

toxic stratus
#

thought I may had to install something on windows like on Linux

#

but I'll try in a new project

toxic stratus
# 
import haxe.Timer;
import db.DatabaseFactory;
class Main {
    
    static function main() {
        trace('here');    
        var timer = new Timer(1000);
        timer.run = () -> {
            trace('running');
        }

        var db = DatabaseFactory.instance.createDatabase(DatabaseFactory.MYSQL, {
            database: 'test',
            host: "127.0.0.1",
            user: "root",
            pass: "123456"
        });

        db.connect().then(function(state) {
            trace('here');
            if (state.data) {
                trace('Database connected');
            } else {
                trace('Database not connected');
            }
        }, (err) -> trace(err));

    }
}

``` just setup a mysql server locally to make sure it isn't some config on the server side
#

this also seemingly fails to connect

#

same code connects on node, weird

toxic stratus
# 
message : Invalid JSON text: "not a JSON text, may need CAST" at position 0 in value for column 'uisettings.bool'.,

I'm getting this error from mysql/dbcore, with the following data:

_record : {
    name : SyncWicks, 
    bool : true
}
#

but true/false is valid json, and if I insert directly into the table it seems to work fine with no cast

#

Not sure where in the chain things are going wrong

velvet lodge
#

could "bool" be reserved?

toxic stratus
toxic stratus
#

named it to flag but have the same issue

toxic stratus
#

this took me quite a bit to hunt down 😭

#

we can't set records to null values!

#

ah, there's empty

#

ahh, but its awkward when working with macros 🤔

#

maybe i'll just set it to an empty string instead

#

Would you be cool if i pr'd setField()?

#

same function just without the null check

#

seems like it messes with the query stuff - gets into weird situations like id=null. Just gonna stick with empty(field)

velvet lodge
#

maybe that field thing is ill concieved, should probably setField and getField

#

its kinda "hidden" behaviour where the function acts a setter depending on what you pass in, which isnt very smart really

toxic stratus
#

it does solve an issue where it won't add empty fields to the list which is convenient

#

sending id = null, where id is a primary key, auto increment causes a mysql syntax error

wide solstice
#

Hello, since I am depending on promises library I need to know which license it has. I want to publish my haxe amqp library to haxelib which is planned to be released under MIT and I guess I've to add a license reference to promises

#

Okay, just saw that it's under MIT, so I've to see how to reference it correctly when publishing it to haxelib 😄

uncut phoenix
wide solstice
#

So for release put both licenses into one file and reference it within haxelib.json?

#

but in license i cannot reference the file, but since both licenses are MIT as far as I see, MIT would be fine right?

wide solstice
#

In case that is fine with just concatenating promises LICENSE and my own LICENSE I can release my amqp library to haxelib 🙂

uncut phoenix
wide solstice
#

Okay, the original promises license remains within the subfolder src/promises. Only thing that is going to be excluded from zip command is the .git folder

uncut phoenix
velvet lodge
#

i know nothing about licences, suffice to say, all the core-haxe stuff is "yours"... i really dont care what you do with it... so if there is a more permissable licence that allows that, just lemme know and ill make that chage (i literally dont care about what you do with the source files, they are "yours" as far as im concerned)

#

(i thought MIT was the most permissive, maybe not?)

toxic stratus
#

I have haxeui ceramic under this license

#

seems like the most anti license, license 😆

uncut phoenix
#

issue with unlicense is that some large companies don't allow their developers to use code under it

#

google is the big one

#

i don't completely understand the reasoning, but MIT is good enough

#

Google does not allow its employees to contribute to projects under public domain equivalent licenses like the Unlicense (and CC0), while allowing contributions to 0BSD licensed and US government PD projects.[6]
from wikipedia

#

fedora doesn't allow unlicensed stuff too

old juniper
# uncut phoenix issue with unlicense is that some large companies don't allow their developers t...

oh... indeed to copy a message https://lists.opensource.org/pipermail/license-review_lists.opensource.org/2012-January/001386.html

Paragraph (and sentence) #1 professes to put the covered work into the
public domain.  As mentioned, paragraph 2 professes to be a grant of
rights normally reserved by default to a copyright owner, which makes no
sense given that the preceding sentence professed to eradicate the
work's quality of being ownable.  _However_ (upon reflection), in itself
that would be harmless if redundant and pointless:  One can interpret
paragraph 2 as an elaboration of the consequences of the first
paragraph.

Paragraph 3 is mostly further explanation of the concept of public
domain, and therefore harmless if not useful.  Its middle sentence 
elaborates that the erstwhile author aims to bind heirs and successors,
too (which is a logical inclusion, irrespective of whether it works).

Paragraph 4, though, is the one that would be amusing if it weren't
tragically broken:  It's the warranty disclaimer.  People accepting the
covered work are obliged to accept the condition of no warranty,
otherwise there is no licence.  Except, oh, wait:  Paragraph 1 professed
to put the work in the public domain, so the erstwhile owner has sawed
off and evaporated in paragraph 1 all power to require the condition in
paragraph 4.
uncut phoenix
#

interesting, kinda what i was putting together

toxic stratus
uncut phoenix
#

0bsd

toxic stratus
#

I'd be surprised if Google wanted to use my haxe projects tho 😄

uncut phoenix
# 
Copyright (C) [year] by [copyright holder] <[email]>

Permission to use, copy, modify, and/or distribute this software for any purpose with or without fee is hereby granted.

THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
#

simple enough

uncut phoenix
toxic stratus
#

issue might be related to my newer haxe/hxcpp versions perhaps

toxic stratus
#

It's haxe version

#

the test code above connects with haxe 4.3.6

#

fails to connect with nightly haxe :/

#

hxcpp git doesn't seem to matter

#

git hxcpp might actually matter if there's haxe version checks

toxic stratus
#

found a nightly haxe version that lets my app compile, can now do some profiling with tracy blobcheer

toxic stratus
#

that's interesting, so core-haxe auto converts Date objects

#

is the timestamp field type supported? I can't recall if i've tried before

#

I can't seem to find any date query api's

#

but it's not a big deal to use the ultra cool raw queries 🙈

velvet lodge
#

when you say date query apis, you mean sql data functions?

toxic stratus
#

my current timestamp column is a double but i'm quickly finding this is not very great for managing queries but I don't know how much compatibility there is for queries around the date column types

velvet lodge
#

cant you just use date objects directly? Just about to head into a meeting, but if that doesnt work, its an api hole - fairly sure i can use less than and greater than in date queries - but maybe i imagined that

toxic stratus
#

I'm not sure, i thought i tried it in the past but maybe what i did was pass a timestamp float to the query instead

toxic stratus
velvet lodge
#

i think they might need to be haxe Date objects

toxic stratus
#

was query sorting added?

velvet lodge
#

... ... i think so, dont remember now, lemme check

#

cant see anything in the unit tests

#

nope - still not implemented im afraid

toxic stratus
#

no worries, thought i remembered something about it being implemented but guess i was remembering wrong

#

have a raw query setup anyway and its easy to switch out whenever its added

velvet lodge
#

thought i remembered something about it being implemented but guess i was remembering wrong
same tbh, but no, nothing there

toxic stratus
#

has the core stack been tested on mac? (ws, http, db)

#

or is it expected to work/not work

toxic stratus
#

http will probably work, ws may be hit or miss but db is the one i'm not so sure about 🤔

velvet lodge
#

i would expect them all to work

#

i dont think they have been tested deeply, although, i have an airbook now, so could (finally)

#

but i think i would expect them to "just work"

#

although, now im thinking about it, i wonder about libmysql

toxic stratus
#

unknown mysql field type (daylow:246), detected as 'NEWDECIMAL' [202000.0000000000] mysql/MySqlClientResult:211

#

it works for nodejs. it's meant to be "the type" to use for currencies

toxic stratus
#

a bit cumbersome to work with on the haxe side tho. seems to be passed around as a string, at least on nodejs hxcpp might be different

#

so got to parseFloat it all the time

#

I suspect this is the mysql db that wants the ssl dllsa

velvet lodge
velvet lodge
# toxic stratus

so yeah, i think it is a dep of mysql, and its not compiled from source (mysqlclient) so its not going to be possible to statically link it either

toxic stratus
# velvet lodge do you have a minimal example? 
CREATE TABLE `testtable` (
    `id` INT NULL,
    `amount` DECIMAL(20,6) NOT NULL
)
COLLATE='utf8mb4_0900_ai_ci';
INSERT INTO `testtable` (`amount`) VALUES (2400.03);

db.table('testtable').then((result) -> {
    return result.table.findOne(query($id == 1));
}).then(function(result) {
  var rec = result.data;
  trace(Type.typeof(rec.field('amount'));
}, (err) -> trace(err));
#

I think that would do it

velvet lodge
#

trace(Type.typeof(rec.field('amount')); says its a string?

toxic stratus
#

for nodejs yes

#

for cpp it doesn't seem to be a mapped field type

velvet lodge
#

not sure what you mean, but its fine, ill put it in a UT and see whats going on

#

my guess is there is an auto conversion case its not hitting

toxic stratus
#

unknown mysql field type (daylow:246), detected as 'NEWDECIMAL' [202000.0000000000] mysql/MySqlClientResult:211
Is the output on cpp

velvet lodge
#

OK, that seems to come from mysql though, which seems be saying its not happy with teh DECIMAL.. ?

toxic stratus
#

maybe the field type is really new?

#

and just not part of the version shipped with dbcore

#

this was what i googled so i just used decimal

velvet lodge
#

sure, im just saying that the error message on hxcpp seems to be coming from mysql, not the db-core stuff

#

... ... actually...

toxic stratus
#

mysql/MySqlClientResult:211 is a core file no?

#

ah lol

#

oh huh

#

ahh there's cpp stuff that has to be done here, thought I might have been able to just add the condition x)

velvet lodge
#

can repro

toxic stratus
#

is it possible to just place the dlls "somewhere" for hxcpp/the app to just detect?

velvet lodge
#

could be put next to the .lib file, but you'd need to copy it to the correct place (via a macro possibly)

toxic stratus
#

would this work cross platform?

velvet lodge
#

nope

toxic stratus
#

or i guess i'd need to find a solution for each os

velvet lodge
#

you would have to have, say, 3 versions and copy appropriately

#

but not sure if that would be ok for linux

#

or if the assumtion could be made that there will always be openssl on linux - not sure

toxic stratus
#

i don't recall needing to install anything ssl on linux

#

don't know about osx

velvet lodge
#

could just be a windows thing then

toxic stratus
#

might have to test that at some point, the awkward thing is all my linux setups are dev environments

#

long term I might be migrating away from mysql in the project anyway and use an api setup instead. But just makes the short term sending to non technical friends to play with a bit awkward

velvet lodge
frosty locust
#

it's not 100% guaranteed that linux would have openssl installed, it's probably safer to have openssl packaged with the binary and to set the executable rpath so it can load it

velvet lodge
frosty locust
#

no, just distributed alongside the binary

velvet lodge
#

right 🙂

frosty locust
#

that would be quite the hack 😅

velvet lodge
#

yeah, i was going to say, would that even work?!? It might need the .dll / .so to even run

frosty locust
#

depends on how it is linked/loaded

velvet lodge
#

yeah

frosty locust
#

if it was with dlopen then you could do that 😅

velvet lodge
frosty locust
#

i think it should be, im not sure of the details of openssl

velvet lodge
#

im actually wondering now if you could statically link openssl into the app and that would be enough, might need to do some shenanigans on the libmysqlclient.h

frosty locust
#

yeah, that would be a simpler solution

velvet lodge
toxic stratus
velvet lodge
toxic stratus
#

a side question, i don't know how this stuff works

#

floats tend to have precision issues

#

in this case is there any chance that an input of Std.parseFloat(22.03) can output something silly like 22.0300000001

#

or will it be 1:1

#

and the precision issues only occur when doing maths

#

i've never seen that occur when doing parsefloat stuff just want some verification

velvet lodge
#

i think it should be fine... there may be a better candidate than float, though i think float is a double in hxcpp, so i think it will be fine (think)

toxic stratus
#

i'll take it

#

i've been using double in mysql and doing the parsefloat stuff everywhere internally anyway

#

issues only seem to occur when doing math ops

#

and then I'll round to decimal places

velvet lodge
#

rounding in crypto can be a dangerous strategy 🙂

toxic stratus
#

i take into account the precision level of the token

velvet lodge
#

yeah, makes sense

toxic stratus
#

and rounding is mostly for surface viewing

#

i'm sure i've made mistakes here and there

velvet lodge
#

the db abstraction is quite "fun" actually 🙂

#

i tend to just use entities2, so forgot about working with the "guts" of db-core

toxic stratus
#

it's nice, i have my own kind of abstraction on top of it

#

it fits in a bit better with my general app flow

velvet lodge
#

heathen!

toxic stratus
#

😄

#

i'll explore e2 one day, just seems unnecessary for now

velvet lodge
#

yeah, its super nice, i only realised the other day that in one project, i was using it with sqlite3 and in another mysql... didnt even really occur to me, doesnt even feel like its database'y... only playing with haxe objects... sort of completely forget they are persisted in the db - and that between two projects a totally different db type

toxic stratus
#

sounds nice

velvet lodge
#

the only difference between the two (apart from the actual entities ofc), is:

EntityManager.instance.database = DatabaseFactory.instance.createDatabase(DatabaseFactory.SQLITE, {
    filename: filename,
    journalMode: journalMode
});
#

and then its just stuff like:

package das.entities.customers;

class CustomerSiteData implements IEntity {
                public var customer:CustomerData;
    @:size(255) public var name:String;
    @:size(255) public var uniqueId:String;
                public var configString:String;
                public var routingSlips:Array<RoutingSlipData>;
                public var initiators:Array<JobInitiatorData>;
    @:cascade   public var accessTokens:Array<CustomerSiteAccessTokenData>;

    ...

    public static function findByUniqueId(uniqueId:String):Promise<CustomerSiteData> {
        return find(Query.query($uniqueId = uniqueId));
    }

    public static function findByCustomer(customer:CustomerData):Promise<Array<CustomerSiteData>> {
        return findAll(Query.query($customer = customer));
    }
}
toxic stratus
# 
Db.getLast('balance', 'aid', Globals.aid, 'timestamp', (record) -> {
  var bal = DBBalance.fromRecord(record);
  trace(bal.tableColumn);
});

Db.get('account', query($aid == Globals.aid && $uid == Globals.user_id), (record) -> {});

Where fromRecord is a macro and these functions just map to haxe enums and get passed to a central database system to get run. A lot of the "fluff" is abstracted away

#

yours seems to go a step further though, by just putting it all on the objects itself

velvet lodge
#

yeah, and creates linking tables, and query caching, etc... its pretty powerful

toxic stratus
#

query caching sounds cool, i have that on my todo list 😅

velvet lodge
#

that the schema it creates, by itself, you literally just concern yourself with the objects, and almost never dive into queries (although, you use the query abstracting in db-core ofc sometimes)

image.png
toxic stratus
#

maybe i'll bump up e2 investigation sooner

#

ah, maybe not, i don't know. I might move the app to an api model and move db to purely server backend

velvet lodge
#

in one app (glint) the db entites are also used as part of an api... which is nice (using core-haxe/seializers)

toxic stratus
#

how does that work?

velvet lodge
#

so you pass objects around that are "db ready"... but entities2 has the concept of "fieldsets" so you can strip certain data

velvet lodge
toxic stratus
#

so many overlapping projects 😩

#

was literally just talking to jeremy about his serialising stuff in tracker lol

velvet lodge
#

the RPC system in glint automatically detects these functions and macros build apis and you just implement the code of the "operation"

#

ill see if i can find a simple exampe, two secs

# 
@:expose @:keep
@:operation(glint.api.users.UsersApi.findUsers)
class FindUsers implements IRpcServerOperation {
    private static var log = new Logger("glint.api.users.server.operations.FindUsers");

    public function execute(criteria:FindUsersCriteria, requestor:UserData):Promise<Array<UserData>> {
        return new Promise((resolve, reject) -> {
            log.info('finding users, criteria: ${criteria} (requestor: ${requestor != null ? requestor.username : null})');

            var queryParts:Array<QueryExpr> = [];
            if (criteria.name != null) {
                queryParts.push(Query.query($username = criteria.name));
            }

            if (criteria.searchTerm != null) {
                queryParts.push(Query.query($username =~ criteria.searchTerm));
            }

            var query = Query.joinQueryParts(queryParts, QBinop.QOpBoolAnd);

            UserData.findAll(query, criteria.pageSize, UserData.NoSensitiveInformation).then(users -> {
                resolve(users);
            }, error -> {
                reject(error);
            });
        });
    }
}
# 
typedef FindUsersCriteria = {
    @:optional public var name:String;
    @:optional public var searchTerm:String;
    @:optional public var pageSize:Null<Int>;
}

interface UsersApi {
    public function findUsers(criteria:FindUsersCriteria, requestor:UserData):Promise<Array<UserData>>;
}
#

then there is an api macro that builds up the client api code (via RPC) so you just go:

Api.instance.users.findUsers(...).then(users -> ...
#

and it makes the round trip to the server, objects are auto serialized / unserialized, etc... and, from the client persective, you just get an array of users... super super simple (lots happening under the hood though ofc)

#

oh, and the rpc operations are all discrete js modules, that are loaded at start up and "auto registered" by the server, which means adding new operations is stupidly simple

toxic stratus
#

what i like about this is the centeralised queries

#

they're all organised in a single place

#

rather than scattered around the app

velvet lodge
#

defo, the idea of entities2 (and 1 really) is that you just dont use queries, i mean, obviously you use db-core queries like above, but you certainly never need insert, update, delete queries etc... you simply just use the objects

#

it feels verrrrry natural (although, ofc, im biased)

velvet lodge
#

@toxic stratus - i dont get that parseFloat issue

# 
    function testBasicSelect(async:Async) {
        var connection:DatabaseConnection = null;
        DBCreator.createConnection("persons2").then(c -> {
            connection = c;
            return connection.query("SELECT * FROM Person");
        }).then(result -> {
            Assert.equals(4, result.data.length);

            Assert.equals(result.data[0].personId, 1);
            Assert.equals(result.data[0].firstName, "Ian");
            Assert.equals(result.data[0].lastName, "Harrigan");
            Assert.equals(result.data[0].iconId, 1);
            Assert.equals(result.data[0].amount, 123.456);

            Assert.equals(result.data[2].personId, 3);
            Assert.equals(result.data[2].firstName, "Tim");
            Assert.equals(result.data[2].lastName, "Mallot");
            Assert.equals(result.data[2].iconId, 2);
            Assert.equals(result.data[2].amount, 222.333);

            connection.close();
            async.done();
        }, error -> {
            trace(error);
            async.done();
        });
    }
#

the .amount stuff is a float, works fine on nodejs

#

(and hxcpp)

toxic stratus
velvet lodge
#

i thought you said you had the issue on nodejs... nodejs doesnt use libmysqlclient

#

it uses mysql2 npm lib

toxic stratus
#

oh my bad, thought it was in core

#

weird 🤔

toxic stratus
velvet lodge
#

could be db-core actually i suppose, just though it might be mysql - quite a few layers here i suppose - ill check that now

velvet lodge
toxic stratus
#

on js "123.456" == 123.456

#

it won't crash

#

it will just "act normally"

#

but the type is actually a string

velvet lodge
#

lol, oh yeah:

            var x:Dynamic = "123.456";
            Assert.equals(x, 123.456);
#

that... erm... seems like a hole in utest

toxic stratus
#

it's a haxe quirk if you ask me lol

velvet lodge
#

i defo would have expected utest to fail there

#

and yeah trace(Type.typeof(result.data[0].amount)); is a string...

#

Ok, thats good then, unit tests were useful to add... sucks that asset didnt pick it up though

#

im trying my hardest not to write a unit test lib

toxic stratus
#

lol

#

haxe has some good support for some unit test libs

#

no idea how it integrates with them all

velvet lodge
#

utest has some quirks im not a fan of tbh

#

that Assert not taking the type into consideration might be the "final straw"

toxic stratus
#

it trips me up so many times in haxe in general

velvet lodge
#

i didnt even know about it

#

like, this is quite laborious to have to do:

            switch (Type.typeof(result.data[0].amount)) {
                case TFloat:
                    // pass    
                case _:    
                    Assert.fail("should be a float");
            }
toxic stratus
#

the incoming is dynamic, it can't really know anything about the type other than what you tell it to be

#

and then its just down to the target differences

velvet lodge
#

yeah, makes sense i guess

#

so really, its js being insane

#

... so it comes back from mysql2 as a string... nice

toxic stratus
#

I think it is stored as a string on mysql's side

#

that might be intentional behaviour

#

to preserve precision

velvet lodge
#

right, makes sense

toxic stratus
#

its just you can still treat it as a number in mysql land

#

it'll get indexed as a number, you can do query maths on it etc

velvet lodge
#

thankfully i already have code that converts js buffers into haxe bytes objects, so just need to expand on that

#

alright, should be fixed now @toxic stratus

toxic stratus
#

nice - thanks!

deep ore
#

oh that's an old message

toxic stratus
#

seems to be tripping up on the json column type now (cpp only - nodejs works fine)

#

public static var JSON:Int = untyped __cpp__("FIELD_TYPE_JSON");
doesn't seem to get picked up

#

./src/mysql/MySqlFieldType.cpp(394): error C2065: 'FIELD_TYPE_JSON': undeclared identifier

velvet lodge
#

my guess is the libmysqlclient being used doesnt work with JSON field types

#

not sure when it was introduced... ... cant find a version in the mysql.h used in the haxe externs either :/

#

... actually, i wonder if ive used the wrong fields:

toxic stratus
#

json fields were added in 5.7+ though

velvet lodge
#

and in fact, the #def version is just using that enum version

toxic stratus
#

oh, i might have it

#

yeah mapping to field type works, but not sure about the type of the rawdata

#

{"lev": 1, "qty": 0.05, "name": "defaults", "tpsl": true, "symbol": "BTCUSDT", "stop_roi": 10, "take_roi": 20, "stop_limit": false, "take_limit": false, "trade_limit":
true}
TBool mysql/MySqlClientResult:193

#

It mapped to bool? 😭

velvet lodge
toxic stratus
# velvet lodge what did you map it to though? And how... really, the lib should be using the en... 
@:headerCode('
#include "mysql.h"
#include "mysql_com.h"
#include "field_types.h"
')
class MySqlFieldType {
        public static var JSON:Int = untyped __cpp__("MYSQL_TYPE_JSON");
    public static var DECIMAL:Int = untyped __cpp__("MYSQL_TYPE_DECIMAL");    
    public static var NEWDECIMAL:Int = untyped __cpp__("MYSQL_TYPE_NEWDECIMAL");    
    public static var TINY:Int = untyped __cpp__("MYSQL_TYPE_TINY");    
    public static var SHORT:Int = untyped __cpp__("MYSQL_TYPE_SHORT");    
    public static var LONG:Int = untyped __cpp__("MYSQL_TYPE_LONG");    
    public static var FLOAT:Int = untyped __cpp__("MYSQL_TYPE_FLOAT");    
    public static var DOUBLE:Int = untyped __cpp__("MYSQL_TYPE_DOUBLE");    
    public static var NULL:Int = untyped __cpp__("MYSQL_TYPE_NULL");    
    public static var TIMESTAMP:Int = untyped __cpp__("MYSQL_TYPE_TIMESTAMP");    
    public static var LONGLONG:Int = untyped __cpp__("MYSQL_TYPE_LONGLONG");    
    public static var INT24:Int = untyped __cpp__("MYSQL_TYPE_INT24");    
    public static var DATE:Int = untyped __cpp__("MYSQL_TYPE_DATE");    
    public static var TIME:Int = untyped __cpp__("MYSQL_TYPE_TIME");    
    public static var DATETIME:Int = untyped __cpp__("MYSQL_TYPE_DATETIME");    
    public static var YEAR:Int = untyped __cpp__("MYSQL_TYPE_YEAR");    
    public static var NEWDATE:Int = untyped __cpp__("MYSQL_TYPE_NEWDATE");    
    public static var ENUM:Int = untyped __cpp__("MYSQL_TYPE_ENUM");    
    public static var SET:Int = untyped __cpp__("MYSQL_TYPE_SET");    
    public static var TINY_BLOB:Int = untyped __cpp__("MYSQL_TYPE_TINY_BLOB");    
    public static var MEDIUM_BLOB:Int = untyped __cpp__("MYSQL_TYPE_MEDIUM_BLOB");    
    public static var LONG_BLOB:Int = untyped __cpp__("MYSQL_TYPE_LONG_BLOB");    
    public static var BLOB:Int = untyped __cpp__("MYSQL_TYPE_BLOB");    
    public static var VAR_STRING:Int = untyped __cpp__("MYSQL_TYPE_VAR_STRING");    
    public static var STRING:Int = untyped __cpp__("MYSQL_TYPE_STRING");    
    public static var CHAR:Int = untyped __cpp__("FIELD_TYPE_CHAR");    
    public static var INTERVAL:Int = untyped __cpp__("FIELD_TYPE_INTERVAL");    
    public static var GEOMETRY:Int = untyped __cpp__("MYSQL_TYPE_GEOMETRY");    
    public static var BIT:Int = untyped __cpp__("MYSQL_TYPE_BIT");    
}
#

added the #include fieldtypes.h

velvet lodge
#

hmmm, i suppose that could work

toxic stratus
#

i'm a complete cpp/cpp extern noob was just seeing if i could fall into a lucky solution x)

velvet lodge
#

... refactoring, 2 secs

velvet lodge
toxic stratus
#

got null errors in my code. trying to figure out source

#

differences between js and cpp 😅

velvet lodge
#

so its working?

toxic stratus
#

i'm not getting unmapped type error anymore

#

just haven't confirmed with actual data input

velvet lodge
#

im parsing the field value too for good measure - but not sure thats wise... i think it is

toxic stratus
#

i think it would be fine as well

#

that's how it works on nodejs

#

and it works okay

velvet lodge
#

cool

toxic stratus
#

issue is a completely different part of my code that i'm using to load the json fields doesn't seem to work at all on cpp

toxic stratus
# 
db.table(table).then((result) -> {
    trace('here');
    trace(Query.queryExprToSql(query));
    return result.table.find(query);
}).then(function(result) {
    trace('here');
    trace(Query.queryExprToSql(query));
    trace(Json.stringify(result.data));
    if (result != null) {
        callback(Records(result.data));
    } else {
        callback(Error('No data', result.data));
    }
}, (err) -> trace(err));
#

this is utter madness 😄

#

One single, simple query, never receives the result from .find only on cpp

#

uid = '0c82d31e-79c1-11ef-96e7-0201461afda4' system/DatabaseSystem:258

#

how the heck am i supposed to repro that 😂

toxic stratus
#

how do I print out the full query again

#

the issue only occurs on cpp so might be some quirk with the query syntax somewhere

#

SELECT * FROM uisettings mysql/MySqlClientConnection:59
WHERE (uisettings.uid = "0c82d31e-79c1-11ef-96e7-0201461afda4")
Nope, query is completely valid

#

yeah i'm stumped, seems like a similar issue to my haxe 4 vs haxe 5 thing, except this is also happening on haxe 4

velvet lodge
#

it doesnt get to the second "here"?

#

what is the table def? is that ui settings json?

toxic stratus
#

CREATE TABLE uisettings (
id INT NOT NULL AUTO_INCREMENT,
uid VARCHAR(50) NOT NULL COLLATE 'utf8mb4_0900_ai_ci',
title VARCHAR(50) NOT NULL DEFAULT 'default' COLLATE 'utf8mb4_0900_ai_ci',
category VARCHAR(50) NOT NULL DEFAULT '0' COLLATE 'utf8mb4_0900_ai_ci',
json JSON NULL DEFAULT NULL,
isdefault INT NOT NULL DEFAULT '0',
number INT NULL DEFAULT NULL,
flag JSON NULL DEFAULT NULL,
string TEXT NULL DEFAULT NULL COLLATE 'utf8mb4_0900_ai_ci',
PRIMARY KEY (id) USING BTREE
)
COLLATE='utf8mb4_0900_ai_ci'
ENGINE=InnoDB
AUTO_INCREMENT=28
;

velvet lodge
#

ill see if i can repro in a UT

toxic stratus
#

oh huh, just noticed i can make isdefault a json and get an actual bool type in the db without the getter 😂

velvet lodge
#

do you delete the db at any point?

#

"drop db ... "

#

i could be totally wrong, and might be something completely seperate, but i seems that in the UT, when i drop the db first, the _handle pointer is no longer valid... ... ... which i think makes sense.. ... ?

toxic stratus
#

oh... drop the db

#

read that as drop the table

#

i don't think so, at least not intentionally. It happens on app init right after connecting

velvet lodge
#

It happens on app init right after connecting
so you do drop the db? Weird it works for you, maybe thats not the issue im seeing and im assuming

toxic stratus
#

i meant the problem occurs right after connecting to the database

#

and other queries go through fine

#

before and after that particular one

velvet lodge
#

DROP DATABASE IF EXISTS persons2;

toxic stratus
#

yeah I defo don't do that anywhere myself

velvet lodge
#

if i put that at the end of my UT, its all fine, if i put that at the start, before creating the db (which should also be fine), then i get errors and it seems the _handle is invalid

#

its probably unlrelated to your issue, but seems like it should be fixed (or at least understood)

#

nvm... im dumb as fuck... the delete code in the unit test also closes the connection... 🤦‍♂️

toxic stratus
#

lol just a consequence of not working with code for a while 😂

velvet lodge
#

i wrote this code like a week ago... 🤫

#

i get an error with this:

# 
connection.query("SELECT * FROM Person WHERE (settings.intSetting= 22)");
# 
Unknown column 'settings.intSetting' in 'where clause'
toxic stratus
#

the table doesn't have a column intSetting

velvet lodge
#

mine does

toxic stratus
#

ahhh

velvet lodge
# 
                promises.push(query.bind(connection, "CREATE TABLE IF NOT EXISTS Person (
                    personId int AUTO_INCREMENT,
                    lastName varchar(50),
                    firstName varchar(50),
                    iconId int,
                    contractDocument blob,
                    amount DECIMAL(20,6),
                    settings json,
                    PRIMARY KEY (personId)
                );"));
toxic stratus
#

am i missing it

velvet lodge
#

missing what?

#

cgpt says this is the correct what to query: SELECT * FROM Person WHERE (settings->'$.intSetting' = 22) and indeed that works

#

i get errors in both hxcpp and nodejs for the "dot" method

toxic stratus
#

ohhh, settings.intSetting is in the json i gotchu

velvet lodge
#

yessir

toxic stratus
#

err, but i'm not querying a json field here

#

i'm querying a regular varchar field

velvet lodge
#

i thought uisettings was a json object... oh well, well, some nice json column unit tests i guess 😄

toxic stratus
#

lmao

velvet lodge
#

i cant repro it at all

toxic stratus
#

yeah, it's a crazy error

#

probably worth not spending time on it

velvet lodge
#

have you tried using backticks in your column names? Its reaching, but maybe its it

toxic stratus
#

in the query string?

#

didn't know that was a thing

velvet lodge
#

in the CREATE TABLE

#

so:

# 
`uid` VARCHAR(50) NOT NULL COLLATE 'utf8mb4_0900_ai_ci'
#

and in the query, you can try a few things:

#

remove the table name, so:

# 
WHERE (uid = '0c82d31e-79c1-11ef-96e7-0201461afda4')"
#

and try and using quote backticks for the column name there too

toxic stratus
#

discord just parsed them so it didn't show

velvet lodge
# 
WHERE (`table`.`column` = 'value')

kinda stuff

toxic stratus
#

doesn't db core do that tho

velvet lodge
velvet lodge
#

ok, so yeah, ive testing this on mysql, maybe its a db-core issue 🤔

toxic stratus
# 
SELECT * FROM `uisettings`
WHERE (`uisettings`.`uid` = "0c82d31e-79c1-11ef-96e7-0201461afda4");
#

the query works in heidi

#

I went to MySqlClientConnection.query and traced the sql param

velvet lodge
#

yeah, so im following now, its db core that built the query, but it seems fine

toxic stratus
#

yeah sorry, i just went digging to try and find where things were going wrong

velvet lodge
#

maybe try a "raw" query there and see if you can whittle it down

toxic stratus
#

oh cool, get to make a cool new Raw(sql) event for my db system 😄

#

even I tried to put off doing that lol

velvet lodge
#

well, tbh, it would be more to understand the issue, if you find a raw query that works, id be interested to know why its different to the generated one

#

i dont personally think there will be a difference... the query looks perfectly valid, works on nodejs for you, and works if you just paste it in

#

so i think something else is happening

#

one thing that feels a little off for me in the generated query is that its using double quotes, i suppose they are interchangable, but it certainly initially look "weird" to me

#

do you happen know ANSI_QUOTES is enabled in your db?

toxic stratus
#

no idea but can find out

toxic stratus
# 
Database connected system/DatabaseSystem:68
SELECT * FROM uisettings WHERE (`uid` = "0c82d31e-79c1-11ef-96e7-0201461afda4"); system/DatabaseSystem:356
Invalid char 0 at position 0 system/DatabaseSystem:357
# 
db.table(table).then((result) -> {
    return result.table.raw(sql);
}).then((result) -> {
    if (result != null) {
        callback(Records(result.data));
    } else {
        callback(Error('No data', result.data));
    }
}, (err) -> {
    trace(sql);
    trace(err);
});
#

Note, that I whittled down the sql to that to try to identify what char 0 is

toxic stratus
#

both work fine in heidi

#

| sql_mode | ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO _DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION |

#

defo disabled

#

found another command I could run

velvet lodge
#

what about:

# 
SELECT * FROM `uisettings` WHERE (`uid` = '0c82d31e-79c1-11ef-96e7-0201461afda4')
toxic stratus
#

Invalid char 0 at position 0 system/DatabaseSystem:357

# 
var query = "SELECT * FROM `uisettings` WHERE (`uid` = '0c82d31e-79c1-11ef-96e7-0201461afda4');";
Db.raw('uisettings', query, (records) -> {});
velvet lodge
#

whats the first param to "raw"?

toxic stratus
velvet lodge
#

what about without the semi colon in the query?

#

(just guessing now)

toxic stratus
#

yea, might not be worth much more effort for now, all other queries seem to work fine

#

it's a strange one

#

looks similar to the no connection on nightly haxe

#

same kind of weirdness

#

yeah ...

SELECT * FROM uisettings system/DatabaseSystem:356
Invalid char 0 at position 0 system/DatabaseSystem:357

toxic stratus
#

issue occurs on linux as well

velvet lodge
#

it has to be something with the db engine you are using... i mean, some setting or something

#

not saying you need to change it... just find out what it is

#

does your remote db have any logging or anything that might give info?

toxic stratus
#

mmm not sure, i'm just using the mysql server that's installed on ubuntu

#

i don't particularly mind changing it if you have a suggestion

velvet lodge
#

i defo dont... just its the only thing remaining i can think of

toxic stratus
#

you know....

#

that error looks like a json error rather than a mysql error

#

very interesting....

#

I deleted all the entries in the table

#

and all of a sudden the query goes through

#

You know what it might be

#

what is a mysql NULL value represented as? Could it be that that is invalid json?

#

i wonder if i set the column default to an empty string if it will all just magically start working

#

json field can't have a default value 🤦‍♂️

velvet lodge
#

ok, that might make sense though, maybe the json parse i added is trying to parse bad json

toxic stratus
velvet lodge
#

although, hang on, this is coming from the sql server

toxic stratus
#

maybe a simple if (data.length != 0) parseJson

#

lemme see if i can test that

velvet lodge
#

the error is coming from the query though

#

the sql server is saying "bad syntax"

toxic stratus
#

is it?

#

but the server accepts the value 🤔

#

could be a bug in the mysql externed tho

velvet lodge
#

its not even getting to the "haxe.Json.parse" part

toxic stratus
#

it's the promise response from the query, i'm not sure where it is in the backend

#

but i'm just assuming here, it looks more like a json error than a mysql error

velvet lodge
#

well, its not in the response processing, im fairly convinced of that (happy to be wrong ofc)

toxic stratus
#

maybe i do have invalid json... lets validate the entries first

#

all entries are valid

#

will now try deleting all the null ones and see if i get a response

velvet lodge
toxic stratus
#

I got rid of all null json values in the db and it works

#

i got my json settings and they seem to be parsed fine

#

will see if it reaches that point with a null entry

toxic stratus
#

repro is easy

#

just add a row of data to your persons2 table with the settings field be set to null

toxic stratus
#

root@ubuntu:~# mysql --version
mysql Ver 8.0.42-0ubuntu0.22.04.1 for Linux on x86_64 ((Ubuntu))

#

so odd

#

oh it has to be an expression! my bad!

noble scaffold
#

Just curious, but why ever deal with the database directly? Do you not like ORMS?

velvet lodge
noble scaffold
#

Is that what NB is using?

uncut phoenix
#

he is using db-core directly

velvet lodge
toxic stratus
#

from what it sounds like its around 10 levels below entities2 from an abstraction pov

uncut phoenix
#

any reason why entities2 isn't just entities lol

velvet lodge
#

but, as an fyi, entities2 is built ontop of db-core

velvet lodge
toxic stratus
#

mostly don't wanna intro it whilst i'm still in the building phase

#

get a better feel for things at this lower level

uncut phoenix
#

you are missing out :p

velvet lodge
uncut phoenix
#

i used it for a small project and it was great

#

the least awful ORM I have used in any language

#

(looking at you gorm)

noble scaffold
#

damn that's a good review

velvet lodge
#

yeah, ive used a bunch, and its defo my fav - i like also that it just works nicely with serialization etc... so i send entities back and forth between client and server (via RPC) and it all "just works"...

#

and no leaking of data because of "fieldsets" (that was one of the issues with entities1)

noble scaffold
#

Might use it for my own stuff then too

toxic stratus
noble scaffold
#

How does it compare to record-macros?

toxic stratus
#

i think record-macros is more comparable to db-core

#

and entities is on steroids

velvet lodge
#

you just deal with objects... entities will take care of all the data, including 1->1 and 1->many relationships

noble scaffold
noble scaffold
toxic stratus
#

backend stuff is generally simple and quick for me to get done

#

but frontend is always a ball ache

velvet lodge
# noble scaffold Oh that's cool

yeah, i played with record macros, and its nice, but its very "manual"... i just generally dont want to think about the db layer anymore - ive written that code 100 times before, and i dont need to write it again 😄

noble scaffold
#

I wonder if you can take that same approach for networking?

toxic stratus
noble scaffold
toxic stratus
#

it kind of just gets in the way

uncut phoenix
#

i should do a weblink+entities2 project someday

velvet lodge
toxic stratus
#

lmao

velvet lodge
#

all using entities, rpc, http, etc

velvet lodge
#

from around there

toxic stratus
#

yeah, i want to be able to "secure" some things a bit better and having a live db connection in app as far as i understand it isn't a great thing x)

#

one less thing to rely on for the client sounds nice in general

velvet lodge
#

db-core already uses connection pooling btw

noble scaffold
#

I think if I hadn't started working on go2hx before this set of libraries came out, I likely would've been a contributor to it, I like the selection

toxic stratus
velvet lodge
velvet lodge
noble scaffold
#

Theoritical is great and all, but production is king.

velvet lodge
#

100%... writing a lib for other people never works out imo... writing a lib for yourself that other people might find useful is the way to go

#

even if no one uses it / them... its perfectly fine

toxic stratus
#

bunch of thoughts scattered around

#

haven't gotten to that point yet

velvet lodge
uncut phoenix
#

i try to make libraries that no one uses

#

including me

velvet lodge
#

haha

uncut phoenix
#

then no one gets hurt

noble scaffold
velvet lodge
toxic stratus
velvet lodge
toxic stratus
#

when you're building something for yourself the priority is the "product/thing"

noble scaffold
toxic stratus
#

and when you're building it for someone else the priority is "returns"

#

maybe not an all encompassing statement to make, but, its certainly an "off" value proposition

velvet lodge
#

i think, for me at least, its about not being able to "fully test"... you are inventing scenarios, non existant usually, its literally creating problems for solutions... with the "its for me" its all about actually making it work for a specific issue... if you can make it a lib and share it across products, even better... and if someone else can glean some use out of it... then even better better

toxic stratus
#

also true, everythings easier when you actually need it

#

the path is clearer

velvet lodge
#

defo, and real use leads to real progress

noble scaffold
velvet lodge
#

entities1 was mostly created as a "im going to have to do some db stuff in a bit, and i dont want to, so ill make a lib"... and it worked, which was great... but when i actually started using it in said product(s), its clear there was design issues (hence entities2)

noble scaffold
#

When you have an exact use in mind and you need it, you can prioritize correctly, when you build for others, it's really hard to map out the priorities because a group of user's wants and needs are constantly fluctuating and it's hard to give the same level of direction

#

I tend to rewrite what I use often 3 times before I am happy with the base design.

#

When I was thinking of just users and not in my own interests, I wasted so much time going down needless rabbit holes and trying to search for problems.

velvet lodge
#

... though, also, tbf, some of the issues in entities1 were just fundemental, like leaking db fields, its just about being naive and coding way to optimistically

noble scaffold
#

I am so much more content working on a problem I know it exists because it's my own and if it indirectly leads to other problems getting fixed that would be great, but letting go of needing/wanting to solve every use case under the sun has been such a relief.

toxic stratus
noble scaffold
#

That sounds golden

toxic stratus
#

the kinds of things that are going on the list change over time, there's less bulky things, less backend stuff, "layer 2" details, oh... those are done, what about "layer 3" ones etc

noble scaffold
velvet lodge
noble scaffold
toxic stratus
#

i don't worry too much about that the time contexts

#

i can just be slow sometimes so i take it into account x)

#

i'll just make a new list and move items over if i get a bit stuck

noble scaffold
#

Yeah it's not a race

toxic stratus
#

dang

#

awkward thing with my little "json" flag trick

#

query($isdefault == true) doesn't bring up any results

#

I'm guessing it likely needs some kind of special syntax

velvet lodge
toxic stratus
#

this might have something to do with it, let me retry "true"

#

yes, $isdefault == "true" works fine when there's actually a true value

#

pretty cool

#

and Type.typeof(settings.isdefault) is bool!

#

this is a neat little setup 😄

velvet lodge
toxic stratus
velvet lodge
#

like, maybe its taking "true" as a string literal and not a bool

velvet lodge
toxic stratus
#

mysql "true" is the equiv of json true