Core Haxe | Haxe | Page 3

velvet lodge Jun 19, 2024, 11:16 AM

#

feels wrong, but alright... either way, i cant add to mysql with just a string value

toxic stratus Jun 19, 2024, 11:17 AM

#

i thought it might have been a modern addition

#

but even haxe.json parses it correctly

velvet lodge Jun 19, 2024, 11:17 AM

#

INSERT INTO persons.person (personId,lastName,firstName,iconId,hourlyRate)
    VALUES (1,'bob','tim',1,1);

#

this fails

#

(firstName is a json column)

#

this passes:

#

INSERT INTO persons.person (personId,lastName,firstName,iconId,hourlyRate)
    VALUES (1,'bob','{}',1,1);

#

("tim" changed to "{}")

toxic stratus Jun 19, 2024, 11:19 AM

#

via db core or mysql gui?

velvet lodge Jun 19, 2024, 11:19 AM

#

gui

#

ok, this is alright though:

INSERT INTO persons.person (personId,lastName,firstName,iconId,hourlyRate)
    VALUES (1,'bob','"tim"',1,1);

So that makes sense seeing as "tim" is valid json but tim (no quotes) isnt

toxic stratus Jun 19, 2024, 11:21 AM

#

velvet lodge ``` INSERT INTO persons.person (personId,lastName,firstName,iconId,hourlyRate) ...

this fails because 'tim' is invalid json

#

change tim to 711 and it will pass

#

if you want it to be valid change it to '"tim"'

velvet lodge Jun 19, 2024, 11:22 AM

#

yeah, i mentioned exactly this above 🙂

toxic stratus Jun 19, 2024, 11:23 AM

#

ah yea

velvet lodge Jun 19, 2024, 11:23 AM

#

ok, so this:

                return result.table.addAll([
                    Person(1, '711', 'Harrigan', 1, Bytes.ofString("this is ians contract document"), 111.222),
                    Person(2, '712', 'Barker', 3, null, 333.444),
                    Person(3, '713', 'Mallot', 2, null, 555.666),
                    Person(4, '714', 'Parker', 1, null, 777.888)
                ]);

and later this:

            var record = result.data;
            record.field("firstName", 777);
            record.field("lastName", "HARRIGAN_MODIFIED");
            record.field("hourlyRate", 999.123);

            return result.table.update(query($personId = 1), record);

works fine

#

so i wonder if we are barking up the wrong tree here and its not that at all

toxic stratus Jun 19, 2024, 11:24 AM

#

could be

#

#

tried single quoting it, and it still fails

velvet lodge Jun 19, 2024, 11:25 AM

#

i think you'll need to minimal repro it and see if it is that... like just a throw away app that opens the db and makes the update

#

is this nodejs btw?

toxic stratus Jun 19, 2024, 11:26 AM

#

yea

velvet lodge Jun 19, 2024, 11:26 AM

#

ok, cool

#

im not convinced it is the json column type now

toxic stratus Jun 19, 2024, 11:27 AM

#

i'm struggling to parse the actual sql error tbh

velvet lodge Jun 19, 2024, 11:27 AM

#

it updates fine for me

toxic stratus Jun 19, 2024, 11:27 AM

#

i can't see what is problematic

velvet lodge Jun 19, 2024, 11:27 AM

#

sql errors are shit in general, especially as the error wont (for good reasons) include the values

#

if it were a problem with the json, i think you would get a CONSTRAINT FAILURE error

#

lets see if because the where is a string value... 👀

#

nope

#

this all works fine:

            var record = result.data;
            record.field("firstName", 777);
            record.field("lastName", "HARRIGAN_MODIFIED");
            record.field("hourlyRate", 999.123);

            var lastName:Dynamic = "Harrigan";
            return result.table.update(query($lastName = lastName), record);

which is about as close as i think i can get to your usage

#

UPDATE Person SET personId = ?, lastName = ?, firstName = ?, iconId = ?, contractDocument = ?, hourlyRate = ? WHERE (`Person`.`lastName` = ?);`

toxic stratus Jun 19, 2024, 11:34 AM

#

velvet lodge this all works fine: ``` var record = result.data; recor...

function test(field:String, value:Dynamic) {
  var record = new Record();
  record.field("firstName", value);
  record.field("lastName", "HARRIGAN_MODIFIED");
  record.field("hourlyRate", 999.123);
  return result.table.update(query($firstName = field), record);
}

test('firstName', 7132);

#

this is pretty much exactly what I have locally

#

some differences

velvet lodge Jun 19, 2024, 11:35 AM

#

i think you'll need to minimal repro it outside of your actual app and see if that makes it make more sense

#

this end, it seems fine

toxic stratus Jun 19, 2024, 11:36 AM

#

velvet lodge this end, it _seems_ fine

can you try the above?

velvet lodge Jun 19, 2024, 11:36 AM

#

whats the difference?

#

its just wrapped in a function

toxic stratus Jun 19, 2024, 11:36 AM

#

you're entering 777 directly in the record instead of having it as a dynamic

#

you're whereing on the last name instead of the first name

velvet lodge Jun 19, 2024, 11:37 AM

#

toxic stratus you're entering `777` directly in the record instead of having it as a dynamic

dont see how that would make a diff, but ill try

toxic stratus Jun 19, 2024, 11:37 AM

#

oh wait, nvm on the last name thing

velvet lodge Jun 19, 2024, 11:37 AM

#

toxic stratus you're whereing on the last name instead of the first name

yours is whering on an unrelated string field though

toxic stratus Jun 19, 2024, 11:38 AM

#

velvet lodge dont see how that would make a diff, but ill try

it's pretty much the only difference i can see between your testcase and my usecase

velvet lodge Jun 19, 2024, 11:38 AM

#

            var record = result.data;
            var value:Dynamic = 777;
            record.field("firstName", value);
            record.field("lastName", "HARRIGAN_MODIFIED");
            record.field("hourlyRate", 999.123);

            var lastName:Dynamic = "Harrigan";
            return result.table.update(query($lastName = lastName), record);

all fine

#

i presume you are using latest everything?

#

i dont think ive fixed anything like this, but who knows

toxic stratus Jun 19, 2024, 11:39 AM

#

yeah just added the libs to hbot yesterday

velvet lodge Jun 19, 2024, 11:40 AM

#

so then, yeah, i guess your only remaning option is to minimal repro it... i cant see another way

#

shouldnt take long, few mins i would have thought

toxic stratus Jun 19, 2024, 11:40 AM

#

velvet lodge ``` var record = result.data; var value:Dynamic = 777; ...

hmmm

#

query($lastName = lastName)

#

in my code I have query($lastName == lastName)

#

is this invalid?

velvet lodge Jun 19, 2024, 11:41 AM

#

oh 😄

#

nope, seems fine

#

yeah, its handled:

                switch (op) {
                    case QOpEq:                 sb.add(" = ");
                    case QOpAssign:             sb.add(" = ");
                }

#

... so its not that... phew... was about to jump off the balcony

toxic stratus Jun 19, 2024, 11:42 AM

#

haha

toxic stratus Jun 19, 2024, 2:45 PM

#

velvet lodge i think you'll need to minimal repro it outside of your actual app and see if th...

i get the same error in a repro project

#

import db.DatabaseError;
import haxe.Timer;
import db.IDatabase;
import db.DatabaseFactory;
import db.Record;
import Query.*;

class Main {
    var db:IDatabase;

    static function main() {
        new Main();
    }

    function new() {
        db = DatabaseFactory.instance.createDatabase(DatabaseFactory.MYSQL, {
            database: "haxe",
            host: "",
            user: "",
            pass: ""
        });

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

        connect();
        var timer = new Timer(500);
        timer.run = () -> {};
    }

    function connect() {
        db.connect().then(function(state) {
            if (state.data) {
                db.table('state').then((result) -> {
                    var record = new Record();
                    record.field("key", "next_roundup");
                    record.field("value", 730);
                    return result.table.update(query($key = "next_roundup"), record);
                }).then(function(res) {
                    trace(res);

                    trace('success');
                    trace(res);
                }, function(err) {
                    trace(err);
                    // trace(queryExprToSql(query));
                    // trace(value.debugString());
                });
                trace('Database connected');
            } else {
                trace('Database not connected');
            }
        },function err(err:DatabaseError) {
                trace(err.call);
                trace(err.message);
        });
    }
}

#

here's my db setup

#

📎 state.sql

#

forgot the table creation

#

lol

velvet lodge Jun 19, 2024, 2:59 PM

#

UPDATE state SET key = ?, value = ? WHERE (`state`.`key` = ?);
TestAll.hx:29: {
        message : You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near 'key = ?, value = 
? WHERE (`state`.`key` = ?)' at line 1,
        call : update

#

🥳

velvet lodge Jun 19, 2024, 3:07 PM

#

toxic stratus i get the same error in a repro project

#

fixed i think... just need to check some other stuff out first

toxic stratus Jun 19, 2024, 3:11 PM

#

yay

velvet lodge Jun 19, 2024, 3:33 PM

#

toxic stratus yay

alright if you pull latest db-core and db-mysql... ... ... ?

toxic stratus Jun 19, 2024, 3:41 PM

#

#

we got something new 😄

velvet lodge Jun 19, 2024, 3:42 PM

#

ok, but that might be "on you" now, no? does your test app work now?

toxic stratus Jun 19, 2024, 3:42 PM

#

Json.stringify(value)

#

boom it works

velvet lodge Jun 19, 2024, 3:42 PM

#

probably needed to be a string or something?

#

although seems odd... didnt need to be in the test app

toxic stratus Jun 19, 2024, 3:43 PM

#

#

in my test app

velvet lodge Jun 19, 2024, 3:43 PM

#

btw, the original issue was nothing to do with json, or anything, its that the "key" column is a reserved word, so needed to be key

toxic stratus Jun 19, 2024, 3:43 PM

#

i get an empty message

toxic stratus Jun 19, 2024, 3:43 PM

#

velvet lodge btw, the original issue was nothing to do with json, or anything, its that the "...

lol

#

you know

#

I thought about that at some point

#

but just let it go as "that can't be it"

velvet lodge Jun 19, 2024, 3:44 PM

#

same issue would have happened for deleting

toxic stratus Jun 20, 2024, 8:12 AM

#

i defo have more normal uses for the mysql db

#

but being able to do something like this in mysql is pretty cool either way

#

it plays really nicely with haxe as well

#

for dbcore the readme says there's one-to-one and one-to-many

#

is many-to-many fine?

velvet lodge Jun 20, 2024, 11:39 AM

#

im not sure ive played with many to many relationships in db-core / entities

toxic stratus Jun 20, 2024, 11:46 AM

#

could something like a tag search work one - many?

#

I don't think it could

toxic stratus Jun 20, 2024, 1:29 PM

#

are LIKE queries supported with db-core?

#

I couldn't find a test for anything that looks like it

velvet lodge Jun 20, 2024, 1:39 PM

#

LIKE isnt supported in query dsl... not sure what the haxe syntax equiv could be

toxic stratus Jun 20, 2024, 1:47 PM

#

you could do an "internally reserved" keyword

#

$LIKE

#

but not sure how elegant that is

velvet lodge Jun 20, 2024, 6:12 PM

#

i wonder if $field % value would be alright?

#

its defo valid syntax, just dunno if its too esoteric

toxic stratus Jun 21, 2024, 8:11 AM

#

might be a bit extra but why not an enum?

#

query(LIKE('a%'));

#

i guess it isn't very universal

velvet lodge Jun 21, 2024, 8:18 AM

#

i just dont like "LIKE"... its very sql... and db-core isnt (supposed) to be sql

#

granted the only two plugins currently are sql, but i dont want to shoot myself in the foot

toxic stratus Jun 21, 2024, 8:32 AM

#

ahhh that makes sense

#

wildcard in itself does seem okay, but it might lack some flexibility

#

like, would i be able to do a%z or a___%

#

oh, i guess so

#

it would probably just look a bit odd

#

$field % a%z

velvet lodge Jun 21, 2024, 8:44 AM

#

i wonder if a whole new operation make sense "search" or something

#

that would mean you cant use "LIKE" in update / delete... but maybe thats ok? Seems dangerous / silly to do things like that anyway?

toxic stratus Jun 21, 2024, 8:48 AM

#

I think that's actually a good compromise

#

because you can still just perform an update or delete with the results of the search altogether

#

I agree it seems a bit odd, but I can also see a use case for these things

#

eg removing duplicates

velvet lodge Jun 21, 2024, 9:03 AM

#

yeah, i can defo see a use for "delete where like"

#

but at the same time, like you say, you could do that with the results of a search

#

im also not sure that "like" would map well to other non-sql things in the future (although "search" almost certainly would)

#

eg, i dont think you could do a "delete where like" on cassandra / solr / lucene (i could be wrong about that ofc)

#

so i guess a seach(fields:Array<String>, terms:Array<String>)? 🤔

#

although how would that work with "(x like a or (y like b and z like c))"... hmm

toxic stratus Jun 21, 2024, 9:10 AM

#

could search accept an expression?

velvet lodge Jun 21, 2024, 9:11 AM

#

you mean a query expression? Like the other calls? Then we are back to "like" in the query dsl

#

maybe a new search dsl? But now its starting so sound over engineered / duplicated

toxic stratus Jun 21, 2024, 9:15 AM

#

ahhh, yeah

#

search(['x like a', 'or y like b'])

#

this would work fine, no?

#

maybe not

#

this type of stuff is complicated 😄

velvet lodge Jun 21, 2024, 9:19 AM

#

toxic stratus `search(['x like a', 'or y like b'])`

thats waaaaay too specific

toxic stratus Jun 25, 2024, 10:45 AM

#

@velvet lodge can I do a direct/inline mysql query with dbcore?

velvet lodge Jun 25, 2024, 10:49 AM

#

nope... i think we discussed it before and consindered adding a "raw" call... ... ?

The thing i dont like about it, is it encourages "just use raw" - though i suppose thats not my business

toxic stratus Jun 25, 2024, 10:50 AM

#

velvet lodge nope... i think we discussed it before and consindered adding a "raw" call... .....

yeah, i vaguely remembered it as well, but discord search within threads sucks

#

its mostly to be able to cover queries that aren't supported in core as of yet

velvet lodge Jun 25, 2024, 10:52 AM

#

yeah, thats fair

toxic stratus Jun 25, 2024, 10:55 AM

#

i would still report them as usual, probably worth putting them up as github issues in general

velvet lodge Jun 26, 2024, 6:34 AM

#

right, you can use "raw" calls now on either ITable or IDatabase... both calls have the same signature (and work in the same way), and you have to "opt in" to allow it:

    #if allow_raw
    public function raw(data:String, values:Array<Any> = null):Promise<DatabaseResult<RecordSet>>;
    #end

#

its not thoroughly tested, so let me know if you get any issues

(side note: "thoroughly" is a bitch of a word to spell - had to google it 😄 )

toxic stratus Jun 27, 2024, 7:53 AM

#

velvet lodge right, you can use "raw" calls now on either ITable or IDatabase... both calls h...

how do we use this?

#

i was expecting something like raw('select foo ...)

velvet lodge Jun 27, 2024, 7:54 AM

#

yeah, thats how you use it, on either ITable or IDatabase... where where you expecting to call "raw" from?

toxic stratus Jun 27, 2024, 7:56 AM

#

velvet lodge yeah, thats how you use it, on either ITable or IDatabase... where where you exp...

oh so data == query

#

what's values for?

velvet lodge Jun 27, 2024, 7:57 AM

#

thats if you want to use it as a prepared statment (which is safer), so you can do:

theDB.raw("select * from x where field = 'bob'")
or
theDB.raw("select * from x where field = ?", ["bob"])

toxic stratus Jun 27, 2024, 7:57 AM

#

ahhhh

velvet lodge Jun 27, 2024, 7:59 AM

#

toxic stratus oh so `data` == `query`

yeah, i didnt want to call it query, because, well, it might not be a query in the future, and in general, i dont like this whole "raw" stuff anyway (hence the opt in)... but i can see the utility in it (to make up for core-db shortcomings... ... ... currently)

toxic stratus Jun 27, 2024, 8:00 AM

#

yeah, i figured its just a better way to handle things when you're busy

velvet lodge Jun 27, 2024, 8:00 AM

#

yeah, makes sense

toxic stratus Jun 27, 2024, 8:00 AM

#

i have no issues changing my code when the proper ways come along :)

velvet lodge Jun 27, 2024, 8:01 AM

#

exactly... it will (hopefully) be a "well, at least i can do stuff until a better method comes along"

toxic stratus Jun 27, 2024, 8:13 AM

#

velvet lodge exactly... it will (hopefully) be a "well, at least i can do _stuff_ until a bet...

#

hell yeah 😄

#

thanks!

velvet lodge Jun 27, 2024, 8:15 AM

#

not something i can be "happy" about as it highlights holes in core-db... but good its not a blocker anymore 🙂

toxic stratus Jun 27, 2024, 8:16 AM

#

src/systems/DatabaseSystem.hx:322: SELECT * FROM quotes WHERE title LIKE '%electr%';
src/systems/DatabaseSystem.hx:61: ceramic electron

#

😄

velvet lodge Jun 27, 2024, 8:16 AM

#

?

toxic stratus Jun 27, 2024, 8:17 AM

#

can search in any position of a string

#

its a luxury that firebase/nosql doesn't provide

toxic stratus Jul 9, 2024, 6:28 PM

#

#

when I insert into the db the column name is id

#

why does it become insertedId here 🤔

velvet lodge Jul 9, 2024, 7:06 PM

#

_insertedId is something that comes from the db when you do an insert, yeah, but it shouldnt have overwritten anything

toxic stratus Jul 9, 2024, 7:07 PM

#

ahh

#

I was expecting the id field to come back as the id field

#

in the insert response

velvet lodge Jul 9, 2024, 7:08 PM

#

i dont think it returns that, unless it requeries it... afaik, the only thing you get back from an insert is the inserted id, i dont really understand why those other fields are there...

#

what is your call?

toxic stratus Jul 9, 2024, 7:09 PM

#

it's the primary key/auto increment field

velvet lodge Jul 9, 2024, 7:09 PM

#

yeah

toxic stratus Jul 9, 2024, 7:09 PM

#

so I just wanted to do stuff with the response and was confused when the id field had no value

#

just a standard insert, this is to do with the response from the insert

velvet lodge Jul 9, 2024, 7:10 PM

#

if you trace the response.data here: https://github.com/core-haxe/db-mysql/blob/main/src/db/mysql/MySqlTable.hx#L158

what do you get?

#

i think i would expect only { insertId: 487 }

#

all that said and done, there is a refreshSchema() call at the start, so technically it should be possible to find out what the primary key is and set that field...

toxic stratus Jul 9, 2024, 7:12 PM

#

db/mysql/MySqlTable.hx:159: {
    fieldCount : 0, 
    affectedRows : 1, 
    insertId : 38, 
    info : , 
    serverStatus : 2, 
    warningStatus : 0, 
    changedRows : 0
}

velvet lodge Jul 9, 2024, 7:13 PM

#

toxic stratus ```hx db/mysql/MySqlTable.hx:159: { fieldCount : 0, affectedRows : 1, ...

so yeah, it doesnt give you the id field name back either

toxic stratus Jul 9, 2024, 7:13 PM

#

nah, its the only field missing

#

seems a bit off

#

considering it does actually send the id

#

just with a different name 😄

velvet lodge Jul 9, 2024, 7:14 PM

#

i think thats usual tbh... ive never seen insert queries be like "this is the primary key name, and its value is 38"

toxic stratus Jul 9, 2024, 7:14 PM

#

possibly, it might legit be over a decade since i used mysql

velvet lodge Jul 9, 2024, 7:15 PM

#

i do like the idea of core-db working it out though... my concern is a) getting it wrong and b) adding overheard to inserts

toxic stratus Jul 9, 2024, 7:16 PM

#

it does seem fitting for core-db

#

the overhead issue, i'm not sure how that could be worked out

velvet lodge Jul 9, 2024, 7:17 PM

#

well, the call to refreshSchema is cached: https://github.com/core-haxe/db-mysql/blob/main/src/db/mysql/MySqlDatabase.hx#L136

#

so its might not be that much of an overheard, but what if the schema changes while things are running... like an ALTER table raw query or something?

#

i guess in that case, since you used "raw" you would have to manualy call clearCachedSchema

toxic stratus Jul 9, 2024, 7:19 PM

#

in this particular instance i'm not using raw

velvet lodge Jul 9, 2024, 7:21 PM

#

no no, i understand, im just saying if you alter table via raw then the cached schema will be out of date, but upon reflection i kinda think "thats a you problem for using raw alter table"... ... if you use addColumn / removeColumn (for example, which will create an alter table query) then it also clears the cached schema

toxic stratus Jul 9, 2024, 7:21 PM

#

if (res.data.hasField('_insertedId')) {
  res.data.field('id', res.data.field('_insertedId'));
}```
simple workaround, just seemed like an oversight

velvet lodge Jul 9, 2024, 7:21 PM

#

well, the problem is knowing its "id"

toxic stratus Jul 9, 2024, 7:22 PM

#

velvet lodge no no, i understand, im just saying _if_ you alter table via raw then the cached...

ahhh, tbh, as long as there's some kind of asterisk somewhere that says "hey these are things to watch out with raw queries, and there may be some other things i've forgotten, its very diy land" etc etc

#

i think it should be okay

#

raw should, for the most part be considered "you're on your own"

velvet lodge Jul 9, 2024, 7:22 PM

#

i think if you use raw, you cant expect db-core to have any idea what you did

toxic stratus Jul 9, 2024, 7:23 PM

#

I would agree

#

can be just "nice" to post some things to look out for, but it should be considered anti db-core

#

and i don't think it should be considered as part of "core's inner workings"

#

at most just some note somewhere in the readme

#

I must say tho

#

having raw around is very comforting x)

velvet lodge Jul 9, 2024, 7:34 PM

#

ask and ye shall receive:

velvet lodge Jul 9, 2024, 7:38 PM

#

toxic stratus having raw around is very comforting x)

its a good thing to have around, but im not a "fan" 🙂

toxic stratus Jul 9, 2024, 8:01 PM

#

velvet lodge ask and ye shall receive: * https://github.com/core-haxe/db-core/commit/a741fa...

oh yay

#

thanks!

toxic stratus Jul 9, 2024, 9:08 PM

#

Hmm, when we do a delete operation

#

is the record we pass to it meant to be used as a kind of where operation

#

or are we meant to supply the full row?

#

ahah

#

okay, i think we have a bit of an unintuitive bug with delete then

#

If the operation fails, there's no clear way to figure out that the operation failed in the response data

#

for example, i purposely supplied a mismatching author_id here, so nothing got deleted from the table

#

but the response is the same as if something did get deleted

velvet lodge Jul 9, 2024, 9:48 PM

#

yeah, sounds like an oversight, ill check it tomorrow... dunno whats better though, returning null result or rejecting (throwing an error)

#

i feel like null result is the best idea, maybe with the property for throwing an exception

toxic stratus Jul 9, 2024, 9:55 PM

#

null or some bool sucess: true, success: false

#

i think error would be somewhat misleading

uncut phoenix Jul 11, 2024, 7:02 PM

#

With core-http's server, is there a way to get the remote IP Address from the request?

toxic stratus Jul 11, 2024, 7:13 PM

#

uncut phoenix With core-http's server, is there a way to get the remote IP Address from the re...

try checking the headers

#

there may be some way to check the ip address via checking the source of the socket request

uncut phoenix Jul 11, 2024, 7:19 PM

#

toxic stratus try checking the headers

nothing in there

velvet lodge Jul 12, 2024, 6:52 AM

#

uncut phoenix With core-http's server, is there a way to get the remote IP Address from the re...

so i dont think it does, shouldnt be hard to add, although, ill have to have a quick check how you can get that info from "normal" haxe http

velvet lodge Jul 12, 2024, 8:45 AM

#

fyi @toxic stratus - im just writing the unit tests, but ive changed the DatabaseResult to include an "itemsAffected" var

#

(rather than success / failure prop)

toxic stratus Jul 12, 2024, 8:46 AM

#

ah neat

#

that's a nice level of granularity

velvet lodge Jul 12, 2024, 8:47 AM

#

yeah, it makes more sense really... so tests like this make sense:

#

    function testBasicDelete_All(async:Async) {
        db.table("Person").then(result -> {
            return result.table.all();
        }).then(result -> {
            Assert.equals(4, result.data.length);
            return result.table.findOne(query($personId = 1));
        }).then(result -> {
            Assert.equals(1, result.data.field("personId"));
            Assert.equals("Ian", result.data.field("firstName"));
            Assert.equals("Harrigan", result.data.field("lastName"));
            Assert.equals(1, result.data.field("iconId"));
            return result.table.deleteAll();
        }).then(result -> {
            Assert.equals(4, result.itemsAffected);
            return result.table.findOne(query($personId = 1));
        }).then(result -> {
            Assert.isNull(result.data);
            return result.table.all();
        }).then(result -> {
            Assert.equals(0, result.data.length);
            async.done();
        }, error -> {
            trace("error", error);
        });
    }

    function testBasicDelete_None(async:Async) {
        db.table("Person").then(result -> {
            return result.table.all();
        }).then(result -> {
            Assert.equals(4, result.data.length);
            return result.table.deleteAll(query($personId = 1111));
        }).then(result -> {
            Assert.equals(0, result.itemsAffected);
            return result.table.all();
        }).then(result -> {
            Assert.equals(4, result.data.length);
            async.done();
        }, error -> {
            trace("error", error);
        });
    }

toxic stratus Jul 12, 2024, 8:48 AM

#

its quite a strong test validation tool in itself

velvet lodge Jul 12, 2024, 8:49 AM

#

with all the core-* libs i tend to unit tests to the nth degree (or try to at least): https://github.com/core-haxe/db-core/tree/main/tests/cases

toxic stratus Jul 12, 2024, 8:52 AM

#

i poke in there when looking to see if there's a feature i want to use now and then 😄

velvet lodge Jul 12, 2024, 8:53 AM

#

PITA is now i have to populate that itemsAffected in all the different routes possible - in just db-sqlite there are 3 routes (npm sqlite3, c++ libsqite3, shitty sys.db)

toxic stratus Jul 12, 2024, 8:56 AM

#

lol you should have a dbcore script that can just update the tables 😂

velvet lodge Jul 12, 2024, 8:57 AM

#

not sure it would work... its more like "add "changes" to SqliteResult", "expose sqlite3_changes call in libsqlite3", "populate chanages from extern call"...

#

not complex, but just alot of places

#

the unit tests help alot ofc.... let me know when ive missed something... which is often 🙈

velvet lodge Jul 12, 2024, 9:19 AM

#

alright, thats in now... you'll need to pull pretty much everything

#

@uncut phoenix - httpRequest now contains remoteAddress property

#

eg:

        httpServer.onRequest = (httpRequest, httpResponse) -> {
            return new Promise((resolve, reject) -> {
                trace(httpRequest.remoteAddress);
                httpResponse.write("this is the response");
                resolve(httpResponse);
            });
        };

toxic stratus Jul 12, 2024, 9:37 AM

#

hell yeah

uncut phoenix Jul 13, 2024, 10:54 PM

#

^ Ah sorry forgot to respond, thanks so much!

velvet lodge Aug 9, 2024, 6:59 PM

#

by [ab]using the OpNegBits unary operator in haxe, we can now have "similar" (aka "LIKE" in sql) queries in the query ast in db-core, eg (from the unit tests):

    function testBasicSimilar_WildCard_Multiple_Or_Nested(async:Async) {
        db.table("Person").then(result -> {
            return result.table.find(query(($firstName =~ "i*" || $firstName =~ "*b") || $firstName = "Jim"));
        }).then(result -> {
            Assert.equals(3, result.data.length);
            Assert.equals("Ian", result.data[0].field("firstName"));
            Assert.equals("Bob", result.data[1].field("firstName"));
            Assert.equals("Jim", result.data[2].field("firstName"));
            async.done();
        }, error -> {
            trace("error", error);
        });
    }

#

this will generate (assuming you are using a sql db-core db plugin):

(`Person`.`firstName` LIKE 'i%' OR `Person`.`firstName` LIKE '%b') OR `Person`.`firstName` = 'Jim'

velvet lodge Aug 28, 2024, 8:04 AM

#

@toxic stratus - what haxe target do you use mysql with?

toxic stratus Aug 28, 2024, 9:03 AM

#

velvet lodge <@151104106973495296> - what haxe target do you use mysql with?

just nodejs atm

#

I think one of core haxe's libs doesn't work on cpp so I'm mostly just using node

velvet lodge Aug 28, 2024, 9:04 AM

#

should all work, i have unit tests

#

but im going to migrate hxcpp to not use sys.db and use libmysqlclient externs (which past me wrote somehow)

toxic stratus Aug 28, 2024, 9:23 AM

#

velvet lodge should all work, i have unit tests

#

🤣

velvet lodge Aug 28, 2024, 9:26 AM

#

heh, thats the core-haxe version, right?

#

but also, thats nothing to do with mysql

#

thats web sockets

toxic stratus Aug 28, 2024, 9:27 AM

#

yeah, but i said one of core-haxe libs doesn't work 😂

velvet lodge Aug 28, 2024, 9:28 AM

#

right, but it does work under nodejs?

toxic stratus Aug 28, 2024, 9:28 AM

#

yeah

velvet lodge Aug 28, 2024, 9:28 AM

#

good to know

toxic stratus Aug 28, 2024, 9:28 AM

#

it is a bit weird tho

#

it has a unit test saying passing on github for hxcpp

velvet lodge Aug 28, 2024, 9:28 AM

#

need to look at core-haxe/ws soon actually as i need it

velvet lodge Aug 28, 2024, 9:28 AM

#

toxic stratus it has a unit test saying passing on github for hxcpp

might not be using wss, just ws

toxic stratus Aug 28, 2024, 9:29 AM

#

ahhh

toxic stratus Aug 28, 2024, 9:29 AM

#

velvet lodge need to look at core-haxe/ws soon actually as i need it

hell yeah

toxic stratus Aug 28, 2024, 9:32 AM

#

velvet lodge <@151104106973495296> - what haxe target do you use mysql with?

what target were you thinking about?

#

i'd like both nodejs/js and cpp available

velvet lodge Aug 28, 2024, 9:33 AM

#

they should all be available now (mysql)

#

but im going to remove the sys.db dep for hxcpp

toxic stratus Aug 28, 2024, 9:39 AM

#

nice, will I need to add any lib for cpp to work?

velvet lodge Aug 28, 2024, 10:00 AM

#

libmysqlclient (which i thought i would have to create, but ive actually already created it)

velvet lodge Aug 28, 2024, 10:43 AM

#

sorry, just reread, currently you wont have to, but after i remove the sys.db dep you will

toxic stratus Aug 28, 2024, 11:08 AM

#

that's cool

toxic stratus Sep 13, 2024, 12:13 PM

#

can't seem to find Mysql after updating db core stuff

#

I installed libmysqlclient

#

oooh

#

i thought libmysqlclient was a complete replacement for hx4compat

#

that fixed it 👍

#

#

lol

#

i'll post a pr I guess 😂

toxic stratus Sep 13, 2024, 8:07 PM

#

message : You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'keys
WHERE (? = ?) LIMIT 1' at line 1,
call : findOne
}

#

query is:

db.table(table).then((result) -> {
  return result.table.findOne(query('aid' == 'idgoeshere'));
})

#

query($aid == 'idgoeshere')) has the same issue

#

SELECT * FROM keys
WHERE (`keys`.`aid` = ?) LIMIT 1;

#

hmm..

#

i wonder if its cause it is "just" keys

toxic stratus Sep 13, 2024, 8:27 PM

#

toxic stratus ```sql SELECT * FROM keys WHERE (`keys`.`aid` = ?) LIMIT 1; ```

this query fails in heidi

#

changing

var sql = 'SELECT ${fieldList} FROM ${table.name}';

to

var sql = 'SELECT ${fieldList} FROM `${table.name}`';

in SqlUtils.hx:53, makes it work but that might breaks things else where

#

lemme know if you're cool with a PR for this change

velvet lodge Sep 14, 2024, 2:02 PM

#

thanks for this btw, merged - accidently left the import in (prolly testing something... no idea)... and so it pulled into the sys.db dep... nice catch 🙂

So is libmysql work for you?

velvet lodge Sep 14, 2024, 2:03 PM

#

toxic stratus changing ```hx var sql = 'SELECT ${fieldList} FROM ${table.name}'; ``` to ```h...

yeah, all names should be quoted, must have missed this one... whats the name of the table? Some reserved name i guess?

toxic stratus Sep 14, 2024, 2:09 PM

#

velvet lodge yeah, all names should be quoted, must have missed this one... whats the name of...

table name was keys

velvet lodge Sep 14, 2024, 2:09 PM

#

mebbe thats reserved in mysql... no idea

#

but yeah, if your fix fixes it, then defo PR please 🙂

toxic stratus Sep 14, 2024, 2:10 PM

#

cools

#

i'm just out of sql for so long no idea if no quoting is relied on behaviour anywhere

velvet lodge Sep 14, 2024, 2:11 PM

#

back ticks is defo the rightway to go

#

every field / table should be backticked

velvet lodge Sep 14, 2024, 2:13 PM

#

toxic stratus i'm just out of sql for so long no idea if no quoting is relied on behaviour any...

unseen core-db issue? "Your sql will get worse since you dont need it anymore" 😄

toxic stratus Sep 14, 2024, 2:14 PM

#

velvet lodge unseen core-db issue? "Your sql will get worse since you dont need it anymore" �...

yeah, there's a few other table names without backticks in here

#

i'll add those in

velvet lodge Sep 14, 2024, 2:14 PM

#

nice one

toxic stratus Sep 14, 2024, 2:14 PM

#

'${table.name}.${tableColumn.name}

#

how should the back ticking be done here?

#

discords markdown is getting in the way here x)

#

velvet lodge Sep 14, 2024, 2:15 PM

#

`table`.`column`

toxic stratus Sep 14, 2024, 2:15 PM

#

this exists, should that be separated?

velvet lodge Sep 14, 2024, 2:16 PM

#

toxic stratus

feels like it yeah

#

might want to runt he UTs just to be sure, but pretty sure thats not the best way to handle these things

#

where is that code?

toxic stratus Sep 14, 2024, 2:17 PM

#

https://github.com/core-haxe/db-core/edit/main/src/db/utils/SqlUtils.hx

#

https://github.com/core-haxe/db-core/pull/5 here's the PR

#

will see what the tests come back as

velvet lodge Sep 14, 2024, 2:18 PM

#

they may fail in general though...

toxic stratus Sep 14, 2024, 2:18 PM

#

ah 😄

velvet lodge Sep 14, 2024, 2:19 PM

#

nodejs and neko should pass though by the looks of it

#

#

yup

#

can you change this one... this one was fine actually, i didnt understand the context... its an "AS"

toxic stratus Sep 14, 2024, 2:21 PM

#

ahhh okay

velvet lodge Sep 14, 2024, 2:21 PM

#

and this one: AS ${joinName}.${tableColumn.name}');

#

any of the "AS"'s

#

just two

#

rest seem fine... quite alot of them... nice catches

toxic stratus Sep 14, 2024, 2:23 PM

#

nice

velvet lodge Sep 14, 2024, 2:24 PM

#

tiny change, but ill make... it will take me longer to explain 😄

#

uh, wait, all the UTs broke locally, might be because of that change i wanted to make, two secs

#

nope, nothing to do with that change, was also backtick issues on db-core/mysql

#

unless you also fixed those?

toxic stratus Sep 14, 2024, 2:30 PM

#

lol what did I fix

#

i thought i undid all the AS'

velvet lodge Sep 14, 2024, 2:30 PM

#

nothing, dont worry... found some other issues

toxic stratus Sep 14, 2024, 2:30 PM

#

ahhh

velvet lodge Sep 14, 2024, 2:30 PM

#

should be fixed now... pull latest db-core / db-mysql to be sure

old juniper Sep 14, 2024, 2:31 PM

#

velvet lodge can you change this one... this one was fine actually, i didnt understand the co...

Hmm ... I think the change here cause problems https://github.com/core-haxe/db-core/commit/000f8535aea3ad68cd508e4cb3098569e690bee2
I had done this change before and https://github.com/core-haxe/db-core/commit/d5c4cbcc5ee7fe11efa017e016707dea8b1f713c ( the first done) I was concentrated on columns and no table names. But I remeber trying what !Billy did and not working

toxic stratus Sep 14, 2024, 2:31 PM

#

oh my bad

#

i didn't remove the ticks from the first half

velvet lodge Sep 14, 2024, 2:32 PM

#

yeah, i fixed those now, but found some issues specific to mysql in the "addColumn" test

#

not sure if they just went under the radar or broke recently... no idea... sorted now and UTs pass locally

#

fyi, these are the commits to get UTs back to normal (locally):

old juniper Sep 14, 2024, 2:36 PM

#

velvet lodge yeah, i fixed those now, but found some issues specific to mysql in the "addColu...

Hmm, one error seems to be fixed but not the one
I have when launching the job-consumer this error. It was the kind of errors when I had when trying to fix the backtick issues

velvet lodge Sep 14, 2024, 2:37 PM

#

sqlite?

toxic stratus Sep 14, 2024, 2:37 PM

#

velvet lodge should be fixed now... pull latest db-core / db-mysql to be sure

runs for me 👍

velvet lodge Sep 14, 2024, 2:37 PM

#

TestAddColumn also fails for me (sqlite)

old juniper Sep 14, 2024, 2:38 PM

#

yeah using sqlite.
I'm nearly sure for some reasons you can't have the backticks in the first part ${table.name}.${tableColumn.name} will try without backticks and see if it works

velvet lodge Sep 14, 2024, 2:40 PM

#

uts pass, just realised you had already made the same change i needed locally (i just hadnt pulled 🙂 )

#

so are you saying with latest db-core, db-sqlite, everything, that exception occurs?

#

(latest as in "3 mins ago")

old juniper Sep 14, 2024, 2:44 PM

#

velvet lodge so are you saying with latest db-core, db-sqlite, everything, that exception occ...

yes 😦

toxic stratus Sep 14, 2024, 2:44 PM

#

🏃‍♂️

velvet lodge Sep 14, 2024, 2:44 PM

#

i feel like something else is up then... or are you saying if you go back some commits then everything works?

old juniper Sep 14, 2024, 2:45 PM

#

velvet lodge i feel like something else is up then... or are you saying if you go back some c...

hmm not sure anymore. I had two problems and one is fixed... and maybe the other is mine... ( hugs !Billy).

velvet lodge Sep 14, 2024, 2:46 PM

#

right, would be good at some point to isolate that... core-db should "just work"... if its a logic error (or whatever) in one of the job consumers... thats a totally seperate thing

#

the UTs certainly pass, but they are only as good as the coverage added by the author (me in this case)

#

lemme check entities real quick, but again, coverage is only as good as "it is" (though coverage on all these libs is pretty good i think)

#

(entities UTs pass - sqlite)

old juniper Sep 14, 2024, 2:51 PM

#

No it doesn't work ( suddenly I thought it didn't work in old versions but it was because of rabbit queues that I forgot to empty) .
Now I'll try with pseudo fix.

velvet lodge Sep 14, 2024, 2:52 PM

#

doesnt work if you go back some commits you mean?

#

rabbit queues that I forgot to empty
this is interesting in its own right, how did you poison your rabbitmq queues? (jeez, had to google how to spell "poison"... :/ )

long agate Sep 14, 2024, 2:55 PM

#

what dbs do u support atm?

velvet lodge Sep 14, 2024, 2:56 PM

#

sqlite / mysql... postgres coming "soon" and "IndexedDB" as a "test of the abstraction"

long agate Sep 14, 2024, 2:57 PM

#

cool i have a custom native sys.db.Postgres impl somewhere, maybe i could butcher that and contribute some code

velvet lodge Sep 14, 2024, 2:58 PM

#

sounds good, for sure... first we need a libpostgres (for hxcpp) but iirc i found it and it looks significantly more sane than libmysqlclient

#

the idea of core-db (not core-haxe in general), is we just stop thinking about "oh, which db"... i guess like haxeui, with the backends and such... im a masochist it seems 😄

long agate Sep 14, 2024, 2:59 PM

#

i used the official dynamic libs via hxcpp i think

old juniper Sep 14, 2024, 2:59 PM

#

velvet lodge doesnt work if you go back some commits you mean?

I'm in stupid mode... I don't know anymore anything ^^ Tell you in a moment when I'm sure what's happening

velvet lodge Sep 14, 2024, 2:59 PM

#

long agate i used the official dynamic libs via hxcpp i think

yeah, i remember looking for libpostgres and it was like libsqlite3... just "there"... libmysqlclient is verrrry different, its embedded in the mysql distro... which is mental

long agate Sep 14, 2024, 3:01 PM

#

sounds weird

velvet lodge Sep 14, 2024, 3:01 PM

#

old juniper I'm in stupid mode... I don't know anymore anything ^^ Tell you in a moment when...

yeah, take your time... its saturday 🙂 ... but there are multiple layers here, and in this context, the layer that is important is db-core... if job consumers are failing and your queues are being poisoned (<---- this fucking word!) then thats a totally different thing, but core-db, at its, erm, core... should function

velvet lodge Sep 14, 2024, 3:02 PM

#

long agate sounds weird

its really odd... like you cant just extract "a dozen" cpp files and be done with it... you need the whole mysql source to get anything to work... which is a cluster fuck

#

so i went with the "precompiled lib" route

#

still need some macro magic on linux tbh to get where headers are etc... but its, effectively, "there"

#

which reminds me that i need to revist that... i set up a VM specifically for it, then went "all done!" (once the VM was setup)... 🤦‍♂️

#

now i look, i wonder how "extractable" libpq is: https://github.com/postgres/postgres/tree/master/src/include/libpq - will find out - id ideally like to use the cpp source and have things statically linked...

old juniper Sep 14, 2024, 3:30 PM

#

velvet lodge right, would be good at some point to isolate that... core-db should "just work"...

Okay it works ^^

velvet lodge Sep 14, 2024, 3:31 PM

#

just about to start the "walk the doggos marathon" but, "everything" is working? Or just core-db (unrelated, but interestingly, i worked it out the other day and i walk 15-18km a day with the 5 dogs in their "groups")

old juniper Sep 14, 2024, 3:34 PM

#

velvet lodge just about to start the "walk the doggos marathon" but, "everything" is working?...

core-db works. There's a bug in migrator, when you have migrator in the db that don't exist.

just about to start the "walk the doggos marathon" but, "everything" is working? Or just core-db (unrelated, but interestingly, i worked it out the other day and i walk 15-18km a day with the 5 dogs in their "groups")
Wow, that's plenty. how many walks do you do a day?

velvet lodge Sep 14, 2024, 3:35 PM

#

core-db works. There's a bug in migrator, when you have migrator in the db that don't exist.
fixable i assume / presume?

Wow, that's plenty. how many walks do you do a day?
3 "groups" (of two), twice a day, about 2-3km a walk

old juniper Sep 14, 2024, 3:37 PM

#

velvet lodge > core-db works. There's a bug in migrator, when you have migrator in the db tha...

fixable i assume / presume?
Workaroundable for now.. but fixable I think 🙂

velvet lodge Sep 14, 2024, 3:37 PM

#

old juniper > fixable i assume / presume? Workaroundable for now.. but fixable I think 🙂

cool

old juniper Sep 14, 2024, 3:38 PM

#

3 "groups" (of two), twice a day, about 2-3km a walk
So 3 to 4 hours of walk ?

velvet lodge Sep 14, 2024, 3:38 PM

#

each walk about 30-45 mins i would say, multiplied by 6 (per day)

#

speed varies on who were taking, if its the two 35kg ones, i cover more distance since they take me for a walk 😄

#

if its the 25kg ones, we tend to amble...

#

when its the tiny one... meh, maybe we dont make it 1km... not sure

old juniper Sep 14, 2024, 3:41 PM

#

velvet lodge each walk about 30-45 mins i would say, multiplied by 6 (per day)

You must be quite healthy, even with the smoking
And you still have room for another dog in your groups 😉

velvet lodge Sep 14, 2024, 3:41 PM

#

You must be quite healthy, even with the smoking
it offsets my smoking quite nicely, for sure

And you still have room for another dog in your groups
🤫 divorces start this way 😄

old juniper Sep 14, 2024, 3:44 PM

#

velvet lodge when its the tiny one... meh, maybe we dont make it 1km... not sure

Oh why, usually tiny dogs like to walk. He's a smeller?

just about to start the "walk the doggos marathon"
Oh have a nice walk!

velvet lodge Sep 14, 2024, 4:54 PM

#

old juniper Oh why, usually tiny dogs like to walk. He's a smeller? > just about to start t...

he hates walking - hes terrified of everything - especially traffic - which we have to pass before we get to the country side

#

(in fact we havent taken him for a while because of that, and the fact he was crazy ill a while back so cant have any vaccinations etc)

toxic stratus Sep 17, 2024, 12:13 PM

#

hmmmm

#

for json field types in db-core

#

hm, nvm, i'm still not actually sure what's going on

#

lemme keep messing aronud

#

I think dce is killing a field but i'm explicitly mentioning it so i'm not sure why 🤔

toxic stratus Sep 17, 2024, 2:04 PM

#

quality of life suggestion

#

when adding an internal id to a record, eg _insertedId, remove them so that I don't have to in my local code 😄

velvet lodge Sep 17, 2024, 2:59 PM

#

i think technically _insertedId isnt needed now... DatabaseResult has properties now for lastInsertedId and affectedRows that dont need to be in the data

toxic stratus Sep 17, 2024, 3:16 PM

#

ohhh nicee

velvet lodge Sep 17, 2024, 3:29 PM

#

yeah, i just need to update a few libs that are dependant on those data level props and get them to use the new result level property

toxic stratus Sep 20, 2024, 8:56 PM

#

hmm, for the websocket api

#

it would be nice that we have a unified state variable

#

I keep getting caught out by trying to send stuff down a connecting socket x)

#

i'm not sure what's going on because my connected var is true from the callback onOpen

#

maybe i've got a bug 🤔

#

yes nvm, was a simple bug that was hidden away x)

toxic stratus Sep 24, 2024, 10:33 AM

#

Hmm, i get weird behaviour with websockets and the websocket server

#

instead of places where a string is expected, I am recieving a UInt8Array

#

socket server I guess I can understand as its still a pure nodejs api

#

but when I send out a string and parse it with a client WebSocket, StrMessage(content) is receiving a UInt8Array

velvet lodge Sep 29, 2024, 6:20 AM

#

@toxic stratus - i think you are using the auto reconnect / replay queries part of db-mysql, right?

#

Im thinking about making couple of tiny changes:

reconnect will be true by default
replay will be true by default
when disconnection detected, it will immediately try and reconnect, not wait for autoReconnectIntervalMS and then try - autoReconnectIntervalMS will come into play if the first reconnection attempt failed

toxic stratus Sep 30, 2024, 9:43 AM

#

velvet lodge Im thinking about making couple of tiny changes: 1) reconnect will be true by d...

I think all of these will be fine for me

#

I see no obvious issues that would occur

velvet lodge Sep 30, 2024, 9:43 AM

#

ive already made the changes and tested locally, all fine

toxic stratus Sep 30, 2024, 10:44 AM

#

https://github.com/core-haxe/db-core/pull/6/files

#

I just posted this to db-core

#

only issue is, i've literally never messed with custom iterators

#

so i'm not sure if there's a "better" way to implement it

#

@velvet lodge

#

but it works either way

velvet lodge Sep 30, 2024, 10:45 AM

#

whats the idea? To get the index and record out?

toxic stratus Sep 30, 2024, 10:45 AM

#

ya

#

key value iterator

#

for (index => value in set)

#

currently you can't get index

velvet lodge Sep 30, 2024, 10:47 AM

#

seems fine to me... not sure why you would need the index exactly, but no doubt very app specific

toxic stratus Sep 30, 2024, 11:03 AM

#

just needed the index for debugging purposes tbh

uncut phoenix Oct 27, 2024, 10:01 PM

#

Does core http have a default timeout? Is there a way to turn this off?

toxic stratus Oct 27, 2024, 10:09 PM

#

I'm pretty sure there is

#

ahh I think I'm remembering delay rather than timeout

uncut phoenix Oct 27, 2024, 10:17 PM

#

I am trying to download a file, but it seems to be closing before it gets the file

velvet lodge Oct 28, 2024, 6:40 AM

#

i dont think it does, unless the underlying impl does... where are you running this? Browser? Got an example url?

uncut phoenix Oct 28, 2024, 11:53 AM

#

velvet lodge i dont think it does, unless the underlying impl does... where are you running t...

try https://fonts.gstatic.com/s/opensans/v40/memSYaGs126MiZpBA-UvWbX2vVnXBbObj2OVZyOOSr4dVJWUgsiH0B4gaVc.ttf

#

i am just doing

client.get(variant.tff).then(r -> {
  sys.io.File.saveBytes(fileName, r.body);
});

toxic stratus Oct 28, 2024, 11:54 AM

#

that's a nice url

uncut phoenix Oct 28, 2024, 11:55 AM

#

thanks, found it myself :p

#

oh and I am using interp target

toxic stratus Oct 28, 2024, 12:16 PM

#

uncut phoenix oh and I am using interp target

try using a different target

toxic stratus Oct 28, 2024, 12:44 PM

#

@uncut phoenix its for debugging sake

uncut phoenix Oct 28, 2024, 12:58 PM

#

Doesn't work with the python target

#

same with js, but I am getting the error:

/home/logo4poopers/Projects/haxe/CeramicFontGrabber/main.js:116
                                request.method = "GET";
                                               ^

TypeError: Cannot set properties of undefined (setting 'method')```

velvet lodge Oct 28, 2024, 1:25 PM

#

uncut phoenix i am just doing ```haxe client.get(variant.tff).then(r -> { sys.io.File.saveBy...

#

package;

import http.HttpClient;

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

        var client = new HttpClient();
        var url = "https://fonts.gstatic.com/s/opensans/v40/memSYaGs126MiZpBA-UvWbX2vVnXBbObj2OVZyOOSr4dVJWUgsiH0B4gaVc.ttf";
        client.get(url).then(r -> {
            trace("got bytes");
            sys.io.File.saveBytes("something.ttf", r.body);
        });
    }
}

#

--class-path src
--main Main

--js bin/main.js

--library hxnodejs
--library http

-D no-deprecation-warnings

--cmd node bin\main.js

#

📎 logo-ttf-bytes-test.zip

uncut phoenix Oct 28, 2024, 1:29 PM

#

velvet lodge ```js package; import http.HttpClient; class Main { static function main()...

Huh, that is what I am doing. Is there some issues relating to running client.get inside another callback?

velvet lodge Oct 28, 2024, 1:30 PM

#

doubtful, you'll have to send a failing sample app - its all fine this end

uncut phoenix Oct 28, 2024, 1:30 PM

#

alright

#

Up to line 24 works fine, then it crashes
https://github.com/l0go/CeramicFontGrabber/blob/main/src/GrabCommand.hx

GitHub

CeramicFontGrabber/src/GrabCommand.hx at main · l0go/CeramicFontGra...

A simple tool for downloading and generating the fonts for haxeui-ceramic - l0go/CeramicFontGrabber

velvet lodge Oct 28, 2024, 1:41 PM

#

i can repro... its a kinda weird way of using promises tbh (i would chain them), but it doesnt really explain the error

toxic stratus Oct 28, 2024, 1:42 PM

#

the url is different, would that matter?

velvet lodge Oct 28, 2024, 1:43 PM

#

yeah, if i hardcode the url to the previous on (https://fonts.gstatic.com/s/opensans/v40/memSYaGs126MiZpBA-UvWbX2vVnXBbObj2OVZyOOSr4dVJWUgsiH0B4gaVc.ttf) then it all works fine

#

could be a redirect that is causing issues - http lib follows redirects, but maybe something is borked there

toxic stratus Oct 28, 2024, 1:44 PM

#

reminds me of an issue i've had before

#

could be a header issue

#

or lack of

velvet lodge Oct 28, 2024, 1:45 PM

#

its not a redirect... its an http, so it should be fine

#

its such a weird error

#

something to do with the HttpRequest abstract

uncut phoenix Oct 28, 2024, 1:48 PM

#

thonk

velvet lodge Oct 28, 2024, 1:49 PM

#

ok, its because variant.tff isnt a string (i think) so it doesnt hit the abstracts "fromString" function

uncut phoenix Oct 28, 2024, 1:49 PM

#

oh, do i need to cast it first?

velvet lodge Oct 28, 2024, 1:50 PM

#

i mean, if you do "" + variant.tff it works, but i dont like it...

#

also, it is a string apparently

#

trace("----------------------------------", variant.ttf, Type.typeof(variant.ttf));

#

src/GrabCommand.hx:26: ----------------------------------, https://fonts.gstatic.com/s/roboto/v32/KFOkCnqEu92Fr1MmgVxIIzc.ttf, TClass({
                __name__ : String
        })

#

i think its a GC issue... i think the variant.tff is being cleaned up super fast... at least, thats all i cant think of

#

because this loop is going exit pretty fast, and "its done" (as i mentioned, its quite weird way to use promises)

uncut phoenix Oct 28, 2024, 1:57 PM

#

would cleaning up the promises fix that?

#

i am not really that familiar with promises (obviously)

velvet lodge Oct 28, 2024, 1:58 PM

#

i just mean that this loop is firing off promises "into the wind"... and maybe by the time the request actually gets round to executing its cleared the mem... its a hunch but its defo possible with this nested promises is loops...

#

gimme a sec, ill show you how i would have chained it (and if the issue persists, ignore all of the above 😄 )

#

well, it wasnt what i thought ;/

#

if i do var s:String = variant.ttf; and the use "s" its all fine

#

but even though variant.ttf is apparently a string, the abstract doesnt think so

#

ah, hang on, ofc it wouldnt... abstracts are compile time

toxic stratus Oct 28, 2024, 2:15 PM

#

is .ttf an abstract over string or a straight string?

velvet lodge Oct 28, 2024, 2:15 PM

#

HttpRequest is an abstract

#

@:forward
@:forward.new
abstract HttpRequest(HttpRequestObject) {
    @:from public static function fromString(s:String):HttpRequest {
        return new HttpRequest(s);
    }
}

#

but it makes perfect sense now... variant.tff is dynamic (at compile time), no way for the compiler to know its a string

toxic stratus Oct 28, 2024, 2:16 PM

#

ahh, yeah that makes sense

velvet lodge Oct 28, 2024, 2:16 PM

#

so yeah, you will need a cast or something like that to let the abstract know its a string

toxic stratus Oct 28, 2024, 2:16 PM

#

if its expected to be a string it should just be typed as one before hand

velvet lodge Oct 28, 2024, 2:17 PM

#

well, it can either be a string, or a HttpRequestObject so you can set headers, etc, etc (hence the abstract)

#

but yeah, it makes perfect sense now i think

#

so yeah @uncut phoenix - you'll need a cast to a Std.string... etc... but fwiw, this is the chaining code (that i dont think you technically need, but its better than firing promises off into the wind an not wait from them to complete):

#

var client = new http.HttpClient();
client.get(API_URL).then(r -> {
    trace("GOT API RESULT");
    var filtered = Fuzzaldrin.filter(r.bodyAsJson, values["family"], {key: "family", maxResults: 1});
    if (filtered.length <= 0) {
        Sys.println("Font family not found: " + values["family"]);
        Sys.exit(-1);
    }

    sys.FileSystem.createDirectory(".tmp");

    return client.get(API_URL + '/${filtered[0].id}');
}).then(r -> {
    trace("GOT FILTERED LIST");
    var promises = [];
    for (variant in (r.bodyAsJson.variants : Array<Dynamic>)) {
        var family = (variant.fontFamily : String).replace(" ", "").replace("'", "");
        var fileName = '.tmp/${family}_${variant.fontWeight}_${styleFormat(variant.fontStyle)}.ttf';
        trace("ADDING FONT TO LIST", variant.ttf + " => " + fileName);
        promises.push(saveFont.bind(client, variant, fileName));
    }

    return PromiseUtils.runAll(promises);
}).then(_ -> {
    trace("ALL DONE!");
}, error -> {
    trace("error", error);
});

...
...

    private function saveFont(client:http.HttpClient, variant:Dynamic, fileName:String):Promise<Bool> {
        return new Promise((resolve, reject) -> {
            trace("MAKING REQUEST", variant.ttf);
            client.get((variant.ttf:String)).then(r -> {
                trace("SAVED BYTES ", fileName, r.body.length);
                sys.io.File.saveBytes(fileName, r.body);
                resolve(true);
            }, error -> {
                reject(error);
            });
        });
    }

toxic stratus Oct 30, 2024, 10:35 AM

#

is there a syntax in Query.query() for limit eg: table.query($uid == 'uid' limit 5)

#

i couldn't find a test for it

#

just seems like findOne exists etc

velvet lodge Oct 30, 2024, 10:38 AM

#

https://github.com/core-haxe/db-core/blob/main/src/db/ITable.hx#L15

toxic stratus Oct 30, 2024, 10:38 AM

#

velvet lodge https://github.com/core-haxe/db-core/blob/main/src/db/ITable.hx#L15

I mean in the query syntax itself

#

probably not, i just woke up not sure if the concept makes sense anymore lol

velvet lodge Oct 30, 2024, 10:43 AM

#

table.page(0, 5, query($uid == 'uid')) will create a limit query

toxic stratus Oct 30, 2024, 10:44 AM

#

ahhh, i was looking at count and was a bit confused lol

#

thanks!

velvet lodge Oct 30, 2024, 10:44 AM

#

.count will create a count(...) query

#

i also think a different name could be nice, well, another call that is a little clearer... "findFixed" or something

#

(but i dont like "fixed")

toxic stratus Oct 30, 2024, 10:45 AM

#

what's wrong with .limit?

velvet lodge Oct 30, 2024, 10:45 AM

#

table.limt(someQuery) doesnt make sense

toxic stratus Oct 30, 2024, 10:46 AM

#

table.results(query, 5) perhaps?

velvet lodge Oct 30, 2024, 10:46 AM

#

also, doesnt seem obvious

#

although, looking at "find" it could have a "maxResults:Null<Int> = null"... ... ?

toxic stratus Oct 30, 2024, 10:47 AM

#

page makes sense now that I look at it fully, but i just scanned the function names and didn't look at the params

velvet lodge Oct 30, 2024, 10:51 AM

#

i think im going to add maxResults also, i think its a useful function param

#

semantically, im not sure "page" always makes sense

toxic stratus Oct 30, 2024, 10:52 AM

#

yeah

#

i skipped it because i just didn't want any "paging"

velvet lodge Oct 30, 2024, 10:53 AM

#

yeah, other times is certainly useful, like when you are actually paging... so the "pageIndex" is nice there

#

but other times, i just want to find(someQuery, 5)

toxic stratus Oct 30, 2024, 10:54 AM

#

yeah, exactly

velvet lodge Oct 30, 2024, 10:55 AM

#

yeah, ill add it as an optional param to "find", and maybe "all"

toxic stratus Oct 30, 2024, 10:58 AM

#

blobthanks

velvet lodge Oct 30, 2024, 10:59 AM

#

np - its nice for my enitites2 lib also

#

MyEntity.findAll(10) for example

#

anoter thing i need to sort out is is sorting

#

possibly another param on find / all

uncut phoenix Oct 30, 2024, 1:45 PM

#

velvet lodge ```js var client = new http.HttpClient(); client.get(API_URL).then(r -> { tr...

Thanks so much! Ffinally having time to look at this. It seems like most of my threads crash tho... Seems like an issue within saveFont

#

i thought it might be smth to do with saveBytes, but commenting that out does the same thing

#

i am assuming it is smth on the core-http level then?

toxic stratus Oct 30, 2024, 1:49 PM

#

uncut phoenix i am assuming it is smth on the core-http level then?

more like just a haxe interplay with dynamics

uncut phoenix Oct 30, 2024, 1:54 PM

#

everything seems to be properly cast tho

velvet lodge Oct 30, 2024, 2:04 PM

#

what does your new code look like?

uncut phoenix Oct 30, 2024, 2:04 PM

#

same as yours

velvet lodge Oct 30, 2024, 2:05 PM

#

is it trace "SAVED BYTES" or... ...

uncut phoenix Oct 30, 2024, 2:05 PM

#

https://gist.github.com/l0go/3e9bfefc3fbd7b622f33f99dc95f2680

#

Yeah, but only for one

velvet lodge Oct 30, 2024, 2:06 PM

#

try return PromiseUtils.runSequentially(promises); , see if that makes a diff

uncut phoenix Oct 30, 2024, 2:07 PM

#

it saves more now

#

like four

#

but still gets stuck on some threads dying

velvet lodge Oct 30, 2024, 2:08 PM

#

this is a sys target, right? Smells like problems with http's internal thread handler

uncut phoenix Oct 30, 2024, 2:08 PM

#

yeah, just interp target

#

guess i should try a different one huh

#

ah yeah, js works

velvet lodge Oct 30, 2024, 2:09 PM

#

i also get the same errors on interp

#

almost certainly an issue with the threading

uncut phoenix Oct 30, 2024, 2:10 PM

#

tbf i might be the only interp user outside of macros lmao

velvet lodge Oct 30, 2024, 2:11 PM

#

yeah, i dont use interp... not sure if there are threading issues there in general

#

lemme try hxcpp just to be sure

#

HL is fine

#

hxcpp also fine

#

ive added a -D http_no_threads flag...

#

which will override even if the target is threaded

#

it does mean that it makes the http lib synchronous though ofc (even though the interface will still be promise based)

#

i actually think it might be that the default http impl in interp is not supposed to be threaded

toxic stratus Oct 30, 2024, 3:04 PM

#

velvet lodge anoter thing i need to sort out is is sorting

could the query expression be "expanded"?

#

just an idea

#

query(expr, limit, sort)

#

i don't think that's too much of an imposition on the abstraction api, is it?

#

and they could just just be ignored paramaters

velvet lodge Oct 30, 2024, 3:14 PM

#

i dont think sort and limit should be part of the query

#

they arent in other db systems (like solr, etc)

#

it also means, if things dont have a sort in them (like, dunno, maybe browser db thing), you can just implement it as a post processing step

#

sort, limit and query should defo be seperate "things"

toxic stratus Oct 30, 2024, 3:31 PM

#

makes sense

toxic stratus Oct 30, 2024, 3:57 PM

#

velvet lodge anoter thing i need to sort out is is sorting

i take it by this, there isn't an api exposed?

#

if not, mind if i expose reverse on recordset?

#

as a temp solution

uncut phoenix Oct 30, 2024, 3:57 PM

#

velvet lodge ive added a `-D http_no_threads` flag...

Now I just get Unix.Unix_error(Unix.EBADF, "recv", "") but i'll use a diff target for now

velvet lodge Oct 30, 2024, 4:05 PM

#

toxic stratus if not, mind if i expose reverse on recordset?

fine by me

toxic stratus Oct 30, 2024, 4:10 PM

#

velvet lodge fine by me

https://github.com/core-haxe/db-core/pull/7 haxethumbsup

weak talon Oct 30, 2024, 4:17 PM

#

uncut phoenix Now I just get ``Unix.Unix_error(Unix.EBADF, "recv", "")`` but i'll use a diff t...

That sounds like the mbedtls issues that should be fixed on nightly.

wide solstice Nov 3, 2024, 3:06 PM

#

Hello, I'm relying heavily on https://github.com/core-haxe/promises for my amqp library to reduce the callback hell necessary for establishing a connection, opening a channel and consuming a queue. I would love to push that library to haxelib but for that I would need promises pushed to haxelib. Is that possible? Also wasn't sure about it, so I created an issue for that

#

Asking because there were bugfixes made and thenshim has been archived in the meantime

velvet lodge Nov 4, 2024, 8:02 AM

#

wide solstice Hello, I'm relying heavily on https://github.com/core-haxe/promises for my amqp ...

hmmmm, yeah, i get the issue - i wasnt going to release any of these as haxelibs tbh - mainly because the names are too generic "http" for example - its also quite alot of burden to keep things up to date - i suppose i could make an exception for promises since its pretty standalone itself (ie, no other core-haxe deps i dont think)

#

i do wonder if there is a better way to handle it though... maybe you could just add its a git module or something and somehow package it with your haxelib? It doesnt change very often at all

wide solstice Nov 4, 2024, 11:46 AM

#

At the moment I am referencing to the library via readme. Packaging it directly via submodule would be another possibility. But I am currently also investigating whether it is possible to add Support for void promises within the typedef

#

Problem with void is that the resolve callback ist typed as Void->Void which makes it Impossible to call it then since Haxe then expects a void Parameter

toxic stratus Nov 4, 2024, 11:54 AM

#

I don't think its intuitive

#

but I think you can do

#

function (_) {}
``` It'll pass

#

i could be remembering wrong there

#

but there was some trick you could do with tink promises for that

wide solstice Nov 4, 2024, 6:14 PM

#

Then the typing would get lost

#

Have to see once I have a bit more time

toxic stratus Nov 10, 2024, 8:19 PM

#

I am thankful we have raw queries in db-core x)

velvet lodge Nov 10, 2024, 8:25 PM

#

toxic stratus I am thankful we have raw queries in db-core x)

boooo - whats the query?

toxic stratus Nov 10, 2024, 8:28 PM

#

velvet lodge boooo - whats the query?

nothing complex

SELECT * FROM `$table` WHERE `$where` = \'$value\' ORDER BY `$column` DESC LIMIT 1

velvet lodge Nov 10, 2024, 8:28 PM

#

right, so sort

toxic stratus Nov 10, 2024, 8:28 PM

#

yeah

#

its just nice that we can continue on

velvet lodge Nov 10, 2024, 8:30 PM

#

yeah, fair enough - i'd actually totally forgotten about all that sort and max results stuff above...

toxic stratus Nov 10, 2024, 8:30 PM

#

completely fine

velvet lodge Nov 10, 2024, 8:31 PM

#

useful additions though, for sure - as you say - raw is "alright" as a stop gap... but if you are using raw, as in this case, it indicates an api hole

wide solstice Nov 11, 2024, 9:06 AM

#

Question about queries, is there a possibility to escape values like you do in PHP with mysqli_real_escape_string?

#

I am planning to use db-core or at least libmysqlclient for my private Project to get rid of libffi

velvet lodge Nov 11, 2024, 9:09 AM

#

if you are using the lower level libs, like libmysqlclient for example, then you probably need to escape things yourself (not sure if there is an exposed function for it in the externs, easy to add if not), if you are using the higher level libs, like db-core, then everything should be auto escaped for you automatically... when the sql is built care is taken for that type of stuff, even identifiers are backticked - and where possible, prepapred statements will be used

wide solstice Nov 11, 2024, 9:10 AM

#

Ah good to know. Have to take a look at libmysqlclient to get it Work on Linux. At the moment the workflows are at least failing

velvet lodge Nov 11, 2024, 9:11 AM

#

ah, would defo be good to know if there are issues - it could just be that the UTs are failing because it can connect to a mysql instance (all that stuff is a PITA)...

#

in fact, actually, i know there arent issues - i know someone who is using it on linux without issues

toxic stratus Nov 11, 2024, 9:12 AM

#

I think the failing might be for a haxe version

velvet lodge Nov 11, 2024, 9:12 AM

#

possibly

toxic stratus Nov 11, 2024, 9:12 AM

#

haxebot uses db core and is running on Linux

wide solstice Nov 11, 2024, 9:13 AM

#

Ah okay, then I have to take a deeper look

velvet lodge Nov 11, 2024, 9:14 AM

#

i would, ofc, suggest you use db-core (ie, the top level abstractions) as, imo (and ofc, im biased) they can save you alot of time, and you can flip db backends at will, etc... and depending on your use case, i would also possibly suggest entities2 (which is, i think, my fav lib, ever)

toxic stratus Nov 11, 2024, 9:14 AM

#

if I recall it might have to be with haxe < 4.3

wide solstice Nov 11, 2024, 9:15 AM

#

Last time I tried libmysqlclient I had compile issues since it couldn't find mysql.h

toxic stratus Nov 11, 2024, 9:16 AM

#

hmm, haxebot runs on nodejs

#

completely forgot that 😅

velvet lodge Nov 11, 2024, 9:16 AM

#

toxic stratus hmm, haxebot runs on nodejs

ah, thats not the same thing then... and actually, now i think about it @old juniper is running on node too

#

so yeah, there still might be issues on *nix

wide solstice Nov 11, 2024, 9:17 AM

#

Also tried to build a Haxe only MySQL connector but failed since it needs Public Key encryption which I didn't want to Dive into 😄

velvet lodge Nov 11, 2024, 9:17 AM

#

i mean, there isnt even anything except a build.windows here: https://github.com/core-haxe/libmysqlclient ... ... so that'll be why 🙂

#

and also this: https://github.com/core-haxe/libmysqlclient/blob/main/src/mysql/RawMySqlClient.hx#L8

#

so yeah, will defo need some rejiggling

wide solstice Nov 11, 2024, 9:19 AM

#

Will have a look at it. Want to get rid of the libffi connector I am using at the moment xD

velvet lodge Nov 11, 2024, 9:21 AM

#

yeah, externs ftw for sure

old juniper Nov 11, 2024, 9:23 AM

#

velvet lodge ah, thats not the same thing then... and actually, now i think about it <@633942...

Hmm yeah running on node right now... but I think I was able to run nearly all tests ( maybe a heaps exception) on dbcore with mysql

velvet lodge Nov 11, 2024, 9:24 AM

#

old juniper Hmm yeah running on node right now... but I think I was able to run nearly all t...

oh yeah... valid point - i think i remember that... but how... there is no linux build xml

#

🤔

#

did you maybe make changes and then not commit them?

old juniper Nov 11, 2024, 9:28 AM

#

velvet lodge did you maybe make changes and then not commit them?

That's absolutely possible... I think I change the code by hand to connect in db, like added a few lines somewhere...

velvet lodge Nov 11, 2024, 9:30 AM

#

yeah, i seem to remember that, and i think i remember a convo about needing a macro that builds the buildxml like hxwidgets does

#

ie, runs some processes to get bits that should be included etc

toxic stratus Nov 11, 2024, 12:27 PM

#

@velvet lodge what's the kind of api you'd want for sorting? something like this find(query:QueryExpr, sort:Direction, allowRelationships:Bool = true)?

#

I can take a shot at implementing it

#

at least for mysql backend

#

oh maybe its more involved than that, we'd need an option to specify a query expression i think 🤔

velvet lodge Nov 11, 2024, 1:12 PM

#

yeah, dont worry, ill do it, i want "maxResults" also

toxic stratus Nov 11, 2024, 1:13 PM

#

no worries, just thought it was low hanging contribution fruit x)

#

thought about it for a bit and realised the abstraction adds quite a few things to consider

wide solstice Nov 12, 2024, 7:49 AM

#

velvet lodge if you are using the lower level libs, like libmysqlclient for example, then you...

But escape of Strings would still be necessary to prevent MySQL injections or are you taking care of it in haxe?

velvet lodge Nov 12, 2024, 7:55 AM

#

in the case of core-db, yeah, its handled internally since it builds the sql queries (you dont write sql in core-db)

wide solstice Nov 12, 2024, 9:35 AM

#

that's what I understood

#

But nevertheless strings need to be escaped by mysql real escape string to prevent sql injections

#

or am I missing something?

velvet lodge Nov 12, 2024, 9:52 AM

#

so, if memory serves, db-core builds up prepared type statements (select from x where field=?) then, it passes that down the stack, some plugins handle it natively (ie, using actual prepared staments) and others (like sys.db which doesnt support prepared statements) builds up the actual sql string and that escapes the values

#

eg, the sys version: https://github.com/core-haxe/mysql/blob/main/src/mysql/impl/sys/DatabaseConnection.hx#L135-L171

#

it will replace the ?s with escaped version (_nativeConnection.escape)

#

nodejs uses the (official?) npm lib, which presumably handles the prepared statement automatically

#

i was looking for the hxcpp version, but it doesnt use prepared statements... though the commit is: remove sys.db dep for hxcpp (1st iteration - not using prepared stmts)

#

and indeed, that is not escaping things, but once it uses prepared statments it shouldnt matter

wide solstice Nov 12, 2024, 10:17 AM

#

ok

#

will have a look at libmysqlclient to get it built with linux. Are you preferring to package the libraries with the repository or would it be okay to depend on installed libmysqlclient?

velvet lodge Nov 12, 2024, 10:45 AM

#

i think you'll need to use whats in the system, the locations can vary so i think we'll need a macro that will build up the @:buildXml dynamically based on system paths (hxWidgets does something similar, but its not a 5 min job)

wide solstice Nov 12, 2024, 10:59 AM

#

Oh, found a way to compile it with linux

#

Extended the build.xml by following:
<section if="linux">
<files id="haxe">
<compilerflag value="-I${include_folder}" />
<compilerflag value="-I${include_folder}/include" />
</files>

    <target id="haxe" tool="linker" toolid="exe">
        <lib name="-lmysqlclient"/>
    </target>
</section>

#

That works at least for haxe tests/common.hxml --class-path tests -cpp target/cpp

#

question is right now, why is the windows build also failing in github workflow

#

sadly I don't have a running windows instance

velvet lodge Nov 12, 2024, 11:03 AM

#

wide solstice question is right now, why is the windows build also failing in github workflow

doesnt have a mysql to connect to

wide solstice Nov 12, 2024, 11:03 AM

#

ah okay

#

then I'll prepare a pull request to extend build.xml by linux and rename it to Build.xml 🙂

velvet lodge Nov 12, 2024, 11:06 AM

#

alot simpler than i thought... might be good to get, dunno, @old juniper to also test

old juniper Nov 12, 2024, 11:13 AM

#

velvet lodge alot simpler than i thought... might be good to get, dunno, <@633942991660384276...

Good look at it later this afternoon

wide solstice Nov 19, 2024, 6:47 PM

#

Good evening, took some time to create the pull request but finally I created it and is ready to merge 🙂

#

build worked on my local machine, don't know whether the github workflows are also executed on pull requests

wide solstice Nov 19, 2024, 7:13 PM

#

https://github.com/core-haxe/libmysqlclient/pull/1

GitHub

Add support for linux target by Dreaded-Gnu · Pull Request #1 · cor...

Renamed build.windows.xml to build.xml
Added support for linux target

toxic stratus Nov 20, 2024, 11:05 AM

#

https://github.com/core-haxe/db-core/pull/8

GitHub

Better error forwarding by Jarrio · Pull Request #8 · core-haxe/db-...

#

did you catch this?

velvet lodge Nov 20, 2024, 12:03 PM

#

yeah, saw both... will merge shortly.. thanks gents! 🙂

uncut phoenix Nov 21, 2024, 7:10 PM

#

in db-core and friends, could we remove dependencies from haxelib.json? I might be using it wrong, but haxelib freaks out when a package that doesn't' exist is added there

toxic stratus Nov 21, 2024, 7:24 PM

#

uncut phoenix in db-core and friends, could we remove dependencies from haxelib.json? I might ...

wdym

uncut phoenix Nov 21, 2024, 7:31 PM

#

toxic stratus wdym

You already have db-core version git installed.
Updating db-core version git ...
db-core was updated
Library db-core current version is now git
Error: Failed with error: No such Project : promises

#

i'm using it through hmm, so maybe it is fine with normal haxelib

toxic stratus Nov 21, 2024, 7:53 PM

#

uncut phoenix ``` You already have db-core version git installed. Updating db-core version git...

ahhh, i think it's better to have it there

#

otherwise it would fail with no error

velvet lodge Nov 21, 2024, 8:00 PM

#

do you have promises installed?

#

i mean, is the error valid?

uncut phoenix Nov 21, 2024, 8:02 PM

#

velvet lodge do you have promises installed?

I guess not at that point, but because I am using hmm I don't really have control on the order of installation

toxic stratus Nov 21, 2024, 8:03 PM

#

uncut phoenix I guess not at that point, but because I am using hmm I don't really have contro...

does hmm have a config file?

#

could try putting promises before db-core

uncut phoenix Nov 21, 2024, 8:04 PM

#

yeah, but i think it gets reset each time you add something new to it

#

not sure if it is intended to be manually edited

#

it sorts alphabetically by default

velvet lodge Nov 21, 2024, 8:05 PM

#

is that an "hmm problem"? I mean, surely dependency resolution is important? I dont think the solution is "remove the depenendancy marker"

#

like, other haxelib.json's have dependencies, surely?

uncut phoenix Nov 21, 2024, 8:06 PM

#

yeah, but they aren't git dependencies

velvet lodge Nov 21, 2024, 8:06 PM

#

ah

uncut phoenix Nov 21, 2024, 8:06 PM

#

anything in dependencies is assumed to be on the haxelib repo

velvet lodge Nov 21, 2024, 8:07 PM

#

the problem is if you remove it form there, you have to manually -lib it in your hxml

#

or maybe you have to anyway 🤔

uncut phoenix Nov 21, 2024, 8:07 PM

#

could you toss them in extraparams?

#

i dont have to right now

velvet lodge Nov 21, 2024, 8:08 PM

#

uncut phoenix could you toss them in extraparams?

i think thats a perfectly fine solution, if it works

#

because, yeah, it pulls them in automatically in vscode, which is useful (not just for building)

#

ie, no promises anywhere explicitly included

uncut phoenix Nov 21, 2024, 8:10 PM

#

Lol, I forgot about your haxeui vscode theme. I can send a PR for extraparams later if it doesn't break anything

velvet lodge Nov 21, 2024, 8:10 PM

#

haxeui vscode theme?

uncut phoenix Nov 21, 2024, 8:10 PM

#

isn't that the haxeui colorscheme?

velvet lodge Nov 21, 2024, 8:11 PM

#

dont think so... i built my own icon pack, thats about it

uncut phoenix Nov 21, 2024, 8:11 PM

#

ah k, i was just looking at the icons

velvet lodge Nov 21, 2024, 8:12 PM

#

yeah, those are intellij icons... i just copy and pasted some icon pack and dropped those icons over the top for choice files

toxic stratus Nov 21, 2024, 9:48 PM

#

uncut phoenix Lol, I forgot about your haxeui vscode theme. I can send a PR for extraparams la...

can we install from git via extraparams? blobthinkingeyes

velvet lodge Nov 21, 2024, 9:49 PM

#

no but you can do --library ...

#

so you'll still get an error etc... but, if it doesnt show up in vscode (as above) then its not good imo

#

i think it should / will

toxic stratus Nov 21, 2024, 9:50 PM

#

velvet lodge no but you can do `--library ...`

I no your no!

#

--L promises:git:https://github.com/core-haxe/promises

#

You can even specify a commit hash if you'd like

velvet lodge Nov 21, 2024, 9:51 PM

#

toxic stratus `--L promises:git:https://github.com/core-haxe/promises`

is that a thing?

toxic stratus Nov 21, 2024, 9:52 PM

#

Yea

velvet lodge Nov 21, 2024, 9:52 PM

#

pretty nice

uncut phoenix Nov 21, 2024, 9:52 PM

#

... can you do that in a haxelib.json

velvet lodge Nov 21, 2024, 9:52 PM

#

but haxelib.json doesnt recognise that?

toxic stratus Nov 21, 2024, 9:52 PM

#

not in the json (at least afaik)

#

but this might solve @wide solstice issue

velvet lodge Nov 21, 2024, 9:53 PM

#

i didnt even know that was a thing... sounds useful for core-haxe, for sure

uncut phoenix Nov 21, 2024, 9:53 PM

#

tried to move all of the libs into extraParams, but I am getting .haxelib/sqlite3/git/src/sqlite/impl/cpp/SqliteDatabase.hx:4: characters 8-21 : Type not found : sqlite.Sqlite

toxic stratus Nov 21, 2024, 9:54 PM

#

you'd need to specify the connector in your project - right?

velvet lodge Nov 21, 2024, 9:54 PM

#

uncut phoenix tried to move all of the libs into extraParams, but I am getting ``.haxelib/sqli...

did you move libmysqlite3 also?

uncut phoenix Nov 21, 2024, 9:55 PM

#

no, i am just doing the stuff in my project rn. which worked fine before. here is the entire diff for that package

diff --git a/extraParams.hxml b/extraParams.hxml
new file mode 100644
index 0000000..d474ee8
--- /dev/null
+++ b/extraParams.hxml
@@ -0,0 +1,4 @@
+# Since core-haxe libraries aren't available on the haxelib repo, we import them here
+-L libsqlite3
+-L promises
+-L logging
diff --git a/haxelib.json b/haxelib.json
index 98f2811..5f402c3 100644
--- a/haxelib.json
+++ b/haxelib.json
@@ -8,13 +8,8 @@
     "tags": [
     ],
     "releasenote": "0.0.0",
-    "dependencies": {
-        "libsqlite3": "",
-        "promises": "",
-        "logging": ""
-    },
     "classPath": "src",
     "name": "sqlite3",
     "description": "",
     "url": ""
-}
\ No newline at end of file
+}

toxic stratus Nov 21, 2024, 9:56 PM

#

i think you're missing a lib

#

sqllite3

#

don't you need that?

uncut phoenix Nov 21, 2024, 9:56 PM

#

this is the sqlite3 lib

toxic stratus Nov 21, 2024, 9:56 PM

#

oh right

toxic stratus Nov 21, 2024, 9:57 PM

#

velvet lodge i didnt even know that was a thing... sounds useful for core-haxe, for sure

https://haxe.org/manual/compiler-usage.html docs if you ever wanna come back to it

velvet lodge Nov 21, 2024, 9:57 PM

#

because of "classPath" i think

#

https://github.com/core-haxe/libsqlite3/blob/main/haxelib.json#L13

uncut phoenix Nov 21, 2024, 9:58 PM

#

oh, how does that affect things?

velvet lodge Nov 21, 2024, 9:58 PM

#

actually, no, nvm, it should be fine... haxelib.json is still there, right?

uncut phoenix Nov 21, 2024, 9:58 PM

#

ye

velvet lodge Nov 21, 2024, 9:58 PM

#

so yeah, ignore me

#

do you have libsqlite3 installed?

uncut phoenix Nov 21, 2024, 10:01 PM

#

yep

velvet lodge Nov 21, 2024, 10:02 PM

#

weird, no idea

uncut phoenix Nov 21, 2024, 10:05 PM

#

i wonder if haxelib does smth fancy

#

looks like libsqlite3 is where sqlite.Sqlite and what not are defined

velvet lodge Nov 21, 2024, 10:13 PM

#

yup

uncut phoenix Nov 21, 2024, 10:18 PM

#

are haxelibs just an abstraction over -cp or do they have their own compilation step?

uncut phoenix Nov 21, 2024, 10:44 PM

#

@frosty locust sorry for pinging, would you happen to have any knowledge in this domain of haxelib?

weak talon Nov 22, 2024, 8:02 AM

#

uncut phoenix are haxelibs just an abstraction over -cp or do they have their own compilation ...

They're just an abstraction over -cp.
You can use haxelib path myLib to check the flags passed to the compiler.

frosty locust Nov 22, 2024, 10:26 AM

#

Yeah, it's just -cp with a couple more compilation flags if the haxelib needs them

wide solstice Nov 22, 2024, 12:42 PM

#

toxic stratus but this might solve <@773097223394558023> issue

I managed to pack promises into my amqp library with some macro setting class path correctly

#

Some question about libmysqlclient: I would like to call real_escape_string, how do I do that / what do I've to do to add a wrapper

velvet lodge Nov 22, 2024, 12:50 PM

#

are you using libmysqlclient directly?

#

oh, i was going to say just use it from here: https://github.com/core-haxe/libmysqlclient/blob/main/src/mysql/MySqlClientConnection.hx

but it doesnt seem to be exposed

#

https://github.com/core-haxe/libmysqlclient/blob/main/src/mysql/RawMySqlClient.hx#L22-L23
externs are there, so really these should be exposed here: https://github.com/core-haxe/libmysqlclient/blob/main/src/mysql/MySqlClientConnection.hx

wide solstice Nov 22, 2024, 12:58 PM

#

ok, will have a look at it.

#

I'm using libmysqlclient directly, yes

#

replaced my libffi library by libmysqlclient

#

just some question: Shouldn't to:RawPointer<cpp.Char> of type ConstCharStar?

#

In case it's correct, how do I pass a string variable in there

velvet lodge Nov 22, 2024, 1:47 PM

#

wide solstice just some question: Shouldn't `to:RawPointer<cpp.Char>` of type `ConstCharStar`?

which function? depends if its supposed to be char* or const char*

wide solstice Nov 22, 2024, 1:47 PM

#

ah okay

wide solstice Nov 22, 2024, 2:24 PM

#

found something in the haxe channel from you regarding the real_escape_string 😄

#

Will play a bit around with that

wide solstice Nov 24, 2024, 8:48 AM

#

What do you think about following code for exposing real_escape_string? Another question shall I also expose real_escape_string_quote?

public function real_escape_string(value:String):String {
    if (_handle == null) {
        throw new Exception("internal handle is null");
    }
    var len = NativeString.utf8Length(value) * 2 + 1;
    var na = NativeArray.create(len);
    var p = Pointer.arrayElem(na, 0);
    RawMySqlClient.real_escape_string(_handle, p.raw, value, value.length);
    return untyped __cpp__("{0}", p.raw);
}

velvet lodge Nov 24, 2024, 8:51 AM

#

wide solstice What do you think about following code for exposing real_escape_string? Another ...

return untyped cpp("{0}", p.raw); does that work?

wide solstice Nov 24, 2024, 8:51 AM

#

looks like it works, still testing around

velvet lodge Nov 24, 2024, 8:51 AM

#

might want to try return new String(untyped cpp("{0}", p.raw)) or something like that, feels safer (could be wrong ofc)

#

also, as for exposing it, i would say that this function should be called "escapeString" and should be here: https://github.com/core-haxe/libmysqlclient/blob/main/src/mysql/MySqlClientConnection.hx ... ... ?

wide solstice Nov 24, 2024, 8:54 AM

#

I put it already into MySqlClientConnection, but will rename it to escapeString

#

will try it with return new string 🙂

velvet lodge Nov 24, 2024, 8:56 AM

#

i could be being over cautious, just not sure what is going to happen to that buffer, might get reclaimed or something

wide solstice Nov 24, 2024, 8:57 AM

#

what about escape string quote?

#

shall I expose that also?

velvet lodge Nov 24, 2024, 8:58 AM

#

could do, wont hurt for it to be there

wide solstice Nov 24, 2024, 9:02 AM

#

ok

#

ahm what about the return of real_escape_string and the quote one? In cpp it returns unsigned long, so shouldn't the return be Int64?

velvet lodge Nov 24, 2024, 9:13 AM

#

probably int will be fine? I mean, technically its not "right" but i doubt anyone is going to be using strings of that size

wide solstice Nov 24, 2024, 9:14 AM

#

ok, just asking because of error handling

#

so how to check for error, when the api states Because mysql_real_escape_string() returns an unsigned value, you can check for -1 by comparing the return value to (unsigned long)-1 (or to (unsigned long)~0, which is equivalent).

#

would uint max value be fine

#

damn arch linux 😄
It's not finding mysql_real_escape_string_quote since the package is mariadb 😒

#

humm... have to see whether a check for existance of mysql_real_escape_string_quote is possible since it's mysql only 😐e

#

at least the mysql_escape_string expose is working 🙂

#

do you want to stick to mysql or mariadb? Seems that linux distributions switched from mysql to mariadb which has a drop in replacement for libmysqlclient but removed a bunch of functions like mysql_real_escape_string_quote

velvet lodge Nov 24, 2024, 9:30 AM

#

i think the lowest common denominator, so imo, just drop mysql_real_escape_string_quote, it was a "nice to have" anyway, mysql_real_escape_string is probably enough... ?

wide solstice Nov 24, 2024, 9:31 AM

#

okay, then I would go and remove the mysql_real_escape_string_quote in favor of mysql_real_escape_string

#

So, I made following modifications:

Change return value of real_escape_string to cpp.Uint64 to be able to check for error
Added escapeString function which exposes real_escape_string
Removed real_escape_string_quote from RawMySqlClient

#

Will do a bunch of further tests with my project and then submit a pull request

#

Translated cpp code looks at the moment like this:

::String MySqlClientConnection_obj::escapeString(::String value){
HX_GC_STACKFRAME(&_hx_pos_63a675c029f1cf5e_88_escapeString)
HXLINE( 89) if (::hx::IsNull( this->_handle )) {
HXLINE( 90) HX_STACK_DO_THROW( : haxe :Exception_obj::_alloc( HX_CTX ,HX("internal handle is null",88,4f,70,9c),null(),null()));
}
HXLINE( 93) int len = ((_hx_utf8_length(value) * 2) + 1);
HXLINE( 95) ::Array< char > na = ::Array_obj< char >::__new(len);
HXLINE( 97) ::cpp::Pointer< char > p = ( (::cpp::Pointer< char >)(::cpp::Pointer_obj::arrayElem(na,0)) );
HXLINE( 99) MYSQL* result = this->_handle;
HXDLIN( 99) char* result1 = p->get_raw();
HXDLIN( 99) ::cpp::UInt64 result2 = ::mysql_real_escape_string(result,result1,value.utf8_str(),value.length);
HXLINE( 101) if ((result2 == ::hx::TCast< ::cpp::UInt64 >::cast(-1))) {
HXLINE( 102) HX_STACK_DO_THROW( : haxe :Exception_obj::_alloc( HX_CTX ,HX("mysql_real_escape_string failed!",0f,53,c1,d0),null(),null()));
}
HXLINE( 105) return ::String(p->get_constRaw());
}

#

which looks valid to me

velvet lodge Nov 24, 2024, 9:39 AM

#

yeah, me too

wide solstice Nov 24, 2024, 9:42 AM

#

only thing which might be incompatible is uint64, because real_escape_string uses normally unsigned long

#

Have to see whether I can reflect that one

#

but generally it seems to work

wide solstice Nov 24, 2024, 10:02 AM

#

got it 😄

#

Used UnsignedLong as return type 😄

#

have to test whether following c++ construct works:
unsigned long result2 = ::mysql_real_escape_string(result,result1,value.utf8_str(),value.length);
if (::hx::IsInstanceEq( result2,::hx::TCast< unsigned long >::cast(-1) )) {
HX_STACK_DO_THROW( : haxe :Exception_obj::_alloc( HX_CTX ,HX("mysql_real_escape_string failed!",0f,53,c1,d0),null(),null()));
}

wide solstice Nov 24, 2024, 10:20 AM

#

looks much better with UnsignedLong return for real_escape_string and untyped __cpp__("unsigned long errorReturn = (unsigned long)-1"); 😄

#

First time working with exposing c++ externs, so sorry for the questions 😄

velvet lodge Nov 24, 2024, 10:23 AM

#

no bother, i didnt actually know about the UnsignedLong type on hxcpp tbh 🙂

wide solstice Nov 24, 2024, 10:23 AM

#

the UnsignedLong was existing in RawMysqlClient.hx
@:native("unsigned long")
extern class UnsignedLong {

}

#

it was used already for some other extern

velvet lodge Nov 24, 2024, 10:24 AM

#

oh, lol, i made it 😄

#

thought it was an hxcpp thing

wide solstice Nov 24, 2024, 10:34 AM

#

hxcpp uses uint64 types which are cross platform

#

sadly libmysqlclient uses unsigned long, so I had to work around that a bit 😄

#

funny, now the github workflows are executing automatically

velvet lodge Nov 24, 2024, 10:39 AM

#

i think they will fail anyway because of mysql not being setup, its a todo list somewhere to sort

wide solstice Nov 24, 2024, 10:51 AM

#

Shall I prepare a pull request for https://github.com/core-haxe/mysql/blob/main/src/mysql/impl/cpp/DatabaseConnection.hx so that it uses the new escapeString method?

GitHub

mysql/src/mysql/impl/cpp/DatabaseConnection.hx at main · core-haxe/...

mysql for all relevant haxe targets. Contribute to core-haxe/mysql development by creating an account on GitHub.

#

or would that be the wrong place within prepareSQL

#

within sys there is an escape call for strings

velvet lodge Nov 24, 2024, 10:52 AM

#

seems sensible, eventually that file will move to prepared statements, but in the meantime, certainly seems safer / apprporiate

wide solstice Nov 24, 2024, 10:52 AM

#

okay, then I'll fork and adjust it 🙂

velvet lodge Nov 24, 2024, 10:52 AM

#

wide solstice okay, then I'll fork and adjust it 🙂

thanks! all very much appreciated 🙂

wide solstice Nov 24, 2024, 10:57 AM

#

do you have any possibility to test the new escape within mysql?

#

Since I'm using libmysqlclient directly and there are no tests I cannot test it

velvet lodge Nov 24, 2024, 10:59 AM

#

yeah, i should be able write UTs for it, the UTs work locally , just not as github actions

wide solstice Nov 24, 2024, 11:03 AM

#

created a pull request that for cpp escapeString of _nativeConnection is used

#

sadly crypto of HaxeFoundation doesn't have a publicEncrypt method. In case it would be there I could have continued my haxe only mysql client 😄

velvet lodge Nov 24, 2024, 11:28 AM

#

tbh, i kinda feel like, for things like this, its usually better to delegate to an external lib... otherwise i think you are always playing catch up

wide solstice Nov 24, 2024, 11:55 AM

#

I guess it depends

#

but yes a cpp mysql extension which is exactly what I need is much easier 😄

toxic stratus Nov 24, 2024, 9:49 PM

#

@velvet lodge @uncut phoenix @wide solstice We CAN add github dependencies via haxelib.json!

uncut phoenix Nov 24, 2024, 9:50 PM

#

ooooo

toxic stratus Nov 24, 2024, 9:50 PM

#

"dependencies": {
  "promises": "git:https://github.com/core-haxe/promises"
}

wide solstice Nov 25, 2024, 7:15 AM

#

Also when I want to publish my package to haxelib? That is why I added promises as submodule with macro as source extending the class-path

velvet lodge Nov 25, 2024, 7:48 AM

#

toxic stratus <@456226577798135808> <@231872730478280705> <@773097223394558023> We CAN add git...

thats pretty nice... i wonder if that will still show up the deps in vscode... @uncut phoenix - does that sort your hmm problem?

toxic stratus Nov 25, 2024, 11:24 AM

#

I haven't tried it with a uploading directly to haxelib

velvet lodge Nov 25, 2024, 11:24 AM

#

these libs arent on haxelib so thats wont be an issue

frosty locust Nov 25, 2024, 11:26 AM

#

Also when I want to publish my package to haxelib?
no, for uploading to haxelib all dependencies need to be on haxelib too

velvet lodge Nov 25, 2024, 11:29 AM

#

ah, sorry, got what NB meant now, thought that was to me

#

so right, Barbaric-Tofu will still need his "copy" solution in place 👌

toxic stratus Nov 25, 2024, 11:54 AM

#

still useful for core haxe I guess tho

velvet lodge Nov 25, 2024, 12:00 PM

#

yeah, be interesting to know if that sorts logo's HMM issue

uncut phoenix Nov 25, 2024, 12:31 PM

#

I am guessing it would. I’ll test it later

toxic stratus Nov 26, 2024, 1:25 PM

#

Does db-core have a syntax for is null or is not null ?

#

oh lol

#

literally the last commit (for tests)

toxic stratus Nov 27, 2024, 12:28 AM

#

https://github.com/core-haxe/http/pull/12/commits

#

I added promises via git to the dependencies here

uncut phoenix Nov 27, 2024, 1:00 AM

#

toxic stratus I added promises via git to the dependencies here

any reason to not do it for logging and queues-core?

toxic stratus Nov 27, 2024, 1:02 AM

#

uncut phoenix any reason to not do it for logging and queues-core?

thanks for volunteering

uncut phoenix Nov 27, 2024, 1:04 AM

#

lol

toxic stratus Nov 27, 2024, 1:06 AM

#

would be quite nice to be able to never need to manually type that url out again tho

velvet lodge Nov 27, 2024, 8:03 AM

#

yeah, i think if this works (as in vscode shows it correctly - which it probably will) then all core-haxe haxelib.jsons should do the same (which ill do at some point)

toxic stratus Dec 1, 2024, 8:19 PM

#

@velvet lodge mysql doesn't work on hxcpp

#

trying to run an application gives the following errors

#

libssl-3-x64.dll was not found
libcrypto-3-x64.dll was not found

#

test case is simple

#

import db.DatabaseFactory;
import db.IDatabase;

class Main {
    static function main() {
        var db:IDatabase = DatabaseFactory.instance.createDatabase(DatabaseFactory.MYSQL, {
            database: "test",
            host: "localhost",
            user: "admin",
            pass: "pass"
        });

        db.connect().then(function(resp) {
            trace('connected');
        }, (err) -> trace(err));
    }
}

velvet lodge Dec 1, 2024, 8:29 PM

#

works fine this end

#

do the UTs also fail for you?

#

#

might need openssl?

toxic stratus Dec 1, 2024, 8:50 PM

#

velvet lodge might need openssl?

oh is this a thing?

#

i thought it would just work

#

#

i do have it somewhere but maybe not installed 🤔

toxic stratus Dec 1, 2024, 8:55 PM

#

velvet lodge might need openssl?

ahh, this was it - thanks!

velvet lodge Dec 1, 2024, 8:59 PM

#

i didnt actually know about the dep, but it makes sense its using ssl... just had it installed so never knew about it

#

i guess that means you'd need to distribute the .dlls with any app shipped? Not sure how i feel about that... would prefer to statically link it

toxic stratus Dec 1, 2024, 9:04 PM

#

yeah it's not very ideal

velvet lodge Dec 1, 2024, 9:05 PM

#

im sure openssl can be used from source and included in with libmysql

#

its a biggun though https://github.com/openssl/openssl

#

looks like it might be able to get away with crypto (https://github.com/openssl/openssl/tree/master/crypto) and ssl (https://github.com/openssl/openssl/tree/master/ssl) which seem simpler and possibly standlone

#

also not sure how to tell libmysql to not use dlls... but there is probably a call spec in the header file somewhere

uncut phoenix Dec 1, 2024, 9:09 PM

#

i wouldn't statically link it on linux

velvet lodge Dec 1, 2024, 9:11 PM

#

its arguably not a great idea to statically link anything "security" like on any system

uncut phoenix Dec 1, 2024, 9:11 PM

#

yeah, hashlink has faced issues with that

velvet lodge Dec 1, 2024, 9:12 PM

#

generally, i prefer statically link (on all OS's) but i think in this case, its a valid exception

uncut phoenix Dec 1, 2024, 9:12 PM

#

for sure

#

plus openssl is huge

velvet lodge Dec 1, 2024, 9:13 PM

#

it is big, defo... and, mysqlclientlib is precompiled too (ie, .lib, not source)

#

might be able to finagle a way to use a statically linked version of openssl, but given the context, i dont think its the right way to go

#

so better to just make the dependency known in the readme or something (i didnt even know about the dep, but makes perfect sense)

uncut phoenix Dec 1, 2024, 9:16 PM

#

you could probably check for it at compile time and give a more in depth error

#

readme good enough for now lol

velvet lodge Dec 1, 2024, 9:16 PM

#

uncut phoenix readme good enough for now lol

yeah, seems overkill when the readme can specify it, the error was pretty clear imo

#

libssl-3-x64.dll was not found
libcrypto-3-x64.dll was not found

uncut phoenix Dec 1, 2024, 9:19 PM

#

yeah

toxic stratus Dec 1, 2024, 9:22 PM

#

velvet lodge so better to just make the dependency known in the readme or something (i didnt ...

yeah, i didn't even consider mysql being the issue

#

cause i saw the tests passing 😂

velvet lodge Dec 10, 2024, 7:56 AM

#

toxic stratus <https://github.com/core-haxe/http/pull/12/commits>

this broke my local build

#

--------------------------------------------------------------------------------
build: glint::analytics::client
--------------------------------------------------------------------------------
 - executing haxe (html5.hxml)
Error: Error: Library promises version git is not installed
     Error: Error: Library promises version git is not installed

#

yup, confirmed, its fine without it... sorry, reverting 🙂

#

issue is that the dep specifies git and i use them as dev, so change is a no-go for me im afraid

toxic stratus Dec 10, 2024, 10:20 AM

#

it's no big deal

#

a gap with haxelib x)

velvet lodge Dec 10, 2024, 10:39 AM

#

yeah, i might look at (but not any time soon) publishing everything to haxelib but under things like core-haxe.http... it would mean -lib would have to change, but actually, -lib core-haxe.http is probably nicer really

knotty vortex Dec 10, 2024, 3:50 PM

#

I'm using

-lib haxeui-blank
--macro haxe.ui.macros.ExternGenerator.generate('src-externs')
--macro include('haxe.ui.components')
--macro include('haxe.ui.containers')
--macro include('haxe.ui.containers.dialogs')
--macro include('haxe.ui.containers.menus')
--macro include('haxe.ui.containers.properties')```
But when compiling against those externs I receive
```src-externs/haxe/ui/backend/ComponentBase.hx:8: characters 124-153 : Not enough type parameters for haxe.ui.core.IEventDispatcher```
Any thoughts?

velvet lodge Dec 10, 2024, 3:55 PM

#

might be a bug in the generator, i defo remember "hand fixing" a few little things

knotty vortex Dec 10, 2024, 4:06 PM

#

ah, thanks

knotty vortex Dec 10, 2024, 4:44 PM

#

For the record to fix after compile:

change haxe.ui.backend.ComponentBase.hx line 8 by changing
extern class ComponentBase extends haxe.ui.backend.ComponentSurface implements haxe.ui.core.IComponentContainer implements haxe.ui.core.IEventDispatcher implements haxe.ui.core.IClonable<haxe.ui.backend.ComponentBase> {
to
extern class ComponentBase extends haxe.ui.backend.ComponentSurface implements haxe.ui.core.IComponentContainer implements haxe.ui.core.IEventDispatcher<haxe.ui.events.UIEvent> implements haxe.ui.core.IClonable<haxe.ui.backend.ComponentBase> {
copied haxe.ui.util.Color from git over extern
copied haxe.ui.validators.Validators.hx from git over extern

Seems to work. Since I copied Color and Validators over extern there may be some file size increase but it was easiest

knotty vortex Dec 10, 2024, 5:51 PM

#

Nope. IDK. Stopping for now with
haxe.ui.containers.TabView is not a constructor ButtonsView ComponentMacros.hx:900
Sorry for the noise

knotty vortex Dec 11, 2024, 5:45 PM

#

Turns out the issue was a line in the xml I was loading. layout="vertical" was throwing issue.
<box width="100%" height="100%" layout="vertical" >
Anyways, it is working. Woohoo. Actually take up more file size, but there it is
https://haxeuitest.github.io/TestApp/html5/index.modular.html

HTML5 Buttons

uncut phoenix Dec 11, 2024, 11:03 PM

#

velvet lodge this broke my local build

How are you managing dependencies right now?

#

just with a hxml?

#

i might just use crobes' haxe gsm

toxic stratus Dec 11, 2024, 11:06 PM

#

i just read gsm as gasm 😄

velvet lodge Dec 12, 2024, 6:20 AM

#

uncut phoenix just with a hxml?

yup, just an hxml

#

actually, im using a haven.xml file, but really that just generates an .hxml

velvet lodge Dec 12, 2024, 6:21 AM

#

knotty vortex Turns out the issue was a line in the xml I was loading. layout="vertical" was...

nice... are you including the whole of haxeui in the module / client?

knotty vortex Dec 12, 2024, 3:30 PM

#

Below is the extern creator
https://github.com/haxeuitest/TestApp/blob/main/html5.GenerateExterns.hxml
This is the Host. Not sure I need the macros in that (tested locally it looks like I do). The DCE that is in here, add/removing had no effect besides smaller.
https://github.com/haxeuitest/TestApp/blob/main/html5.host.hxml
This is the Client. I just converted the box with layout vertical into vbox for now
https://github.com/haxeuitest/TestApp/blob/main/html5.hxml

GitHub

TestApp/html5.GenerateExterns.hxml at main · haxeuitest/TestApp

Contribute to haxeuitest/TestApp development by creating an account on GitHub.

GitHub

TestApp/html5.host.hxml at main · haxeuitest/TestApp

Contribute to haxeuitest/TestApp development by creating an account on GitHub.

GitHub

TestApp/html5.hxml at main · haxeuitest/TestApp

Contribute to haxeuitest/TestApp development by creating an account on GitHub.

#

You can see it working here https://haxeuitest.github.io/TestApp/html5/index.modular.html
the buttons, labels buttons are in the host

HTML5 Buttons

old juniper Dec 12, 2024, 3:55 PM

#

knotty vortex Below is the extern creator https://github.com/haxeuitest/TestApp/blob/main/html...

Cool 🙂 Just curious why in core haxe and not in haxeui? Is it using some core haxe stuff?

knotty vortex Dec 12, 2024, 4:04 PM

#

old juniper Cool 🙂 Just curious why in core haxe and not in haxeui? Is it using some core ...

Discussion was intended to focus on modular from core haxe which is outside of HaxeUI sphere.

velvet lodge Dec 13, 2024, 8:26 AM

#

So yeah, if you include all of haxeui in the modular host (https://github.com/haxeuitest/TestApp/blob/main/html5.host.hxml#L11-L15), then you wont see any particular size decrease - in fact, as you noticed, you may see a since increase since you are including everything. The idea would be to only include the classes you want in the modular host... I have some other ideas about including the externs in host and the concrete impls in modules (that would be loaded dynamically), i already do that for the "entities" in the glint application im writing, ie, the host has no clue about the the "enitites", and other modules use them via externs, but once the entities module is loaded, it provides the impls at runtime

#

(there is no reason why haxeui component impls - or anything really - couldnt work the same way)

knotty vortex Dec 13, 2024, 10:28 PM

#

makes sense. My main goal was to allow compile times to decrease. Reducing host would likely mean more compile times? Sounds good though. Defs nice for release.

toxic stratus Dec 16, 2024, 5:20 PM

#

Yo Ian, I have some info regarding websockets

#

i've managed to connect to a socket via wss

#

on native/hxcpp

#

I had to change this line from sleep(0) to sleep(0.1) https://github.com/core-haxe/websockets/blob/main/src/ws/impl/sys/WebSocket.hx#L148

#

it was completely blocking my app

#

that's the "only" thing I had to do really, so i'm not sure how valid that change is or complete, but for some reason ws isn't working and errors out with an error SSL - An invalid SSL record was received

#

even though it's meant to just be a local socket connection

velvet lodge Dec 16, 2024, 5:30 PM

#

toxic stratus I had to change this line from `sleep(0)` to `sleep(0.1)` <https://github.com/co...

hmmmm

#

its probably a valid change, but i think that stuff need work anyway, this is still "half ported"

toxic stratus Dec 16, 2024, 5:31 PM

#

just thought i'd mention it because i think this was on your todo list

velvet lodge Dec 16, 2024, 5:31 PM

#

not sure about the wss though, i would have expected it to "just fail"

toxic stratus Dec 16, 2024, 5:33 PM

#

velvet lodge not sure about the wss though, i would have expected it to "just fail"

import haxe.Json;
import ws.WebSocket;

class Main {
    static function main() {
        trace("Hello, world!");
        var ws = new WebSocket('wss://stream.bybit.com/v5/public/spot');
        ws.onopen = () -> {
            trace('opened');
            ws.send(Json.stringify({
                req_id: "test", // optional
                op: "subscribe",
                args: ["kline.1.BTCUSDT"]
            }));
        }

        ws.onerror = (errr) -> {
            trace(errr);
        }

        ws.onmessage = (message) -> {
            trace(message);
        }
        ws.onclose = () -> {
            trace('socket closed');
        }
    }
}
``` here's a testcase if you want to visit it at some point

#

might require git hxcpp i'm not sure

#

it's also interesting to note, i only need to adjust the sleep for ceramic

velvet lodge Dec 16, 2024, 5:34 PM

#

toxic stratus ```hx import haxe.Json; import ws.WebSocket; class Main { static function m...

https://github.com/core-haxe/websockets/blob/main/src/ws/impl/sys/WebSocket.hx#L66 - i dont know why that doesnt just throw this exception

toxic stratus Dec 16, 2024, 5:34 PM

#

📎 WebSocket.hx

velvet lodge Dec 16, 2024, 5:34 PM

#

toxic stratus it's also interesting to note, i only need to adjust the sleep for ceramic

ah, it probably screws with the MainLoop

toxic stratus Dec 16, 2024, 5:34 PM

#

here's my modded websockets file

velvet lodge Dec 16, 2024, 5:34 PM

#

ah, you removed the throw

toxic stratus Dec 16, 2024, 5:34 PM

#

makes one other change (removes the throw for not implemented)

velvet lodge Dec 16, 2024, 5:35 PM

#

but the throw was there because its not implemented / tested - so im not surprised it didnt work 🙂

toxic stratus Dec 16, 2024, 5:35 PM

#

but it does work!

velvet lodge Dec 16, 2024, 5:35 PM

#

except it didnt right? It failed the SSL handshake

toxic stratus Dec 16, 2024, 5:35 PM

#

for ws

#

not wss

velvet lodge Dec 16, 2024, 5:36 PM

#

so, hang on, wss worked and ws failed an SSL handshake?

toxic stratus Dec 16, 2024, 5:36 PM

#

it's backwards 😄

toxic stratus Dec 16, 2024, 5:36 PM

#

velvet lodge so, hang on, wss worked and ws failed an SSL handshake?

yes!

velvet lodge Dec 16, 2024, 5:36 PM

#

very odd

toxic stratus Dec 16, 2024, 5:37 PM

#

i've got a local websocket server websockets one

#

afaik that requires a ws connection

velvet lodge Dec 16, 2024, 5:37 PM

#

right, ill have to play with this and get some UTs going

toxic stratus Dec 16, 2024, 5:38 PM

#

works on nodejs but fails with SSL - An invalid SSL record was received

#

on cpp

velvet lodge Dec 16, 2024, 5:38 PM

#

and this is ws, not wss?

toxic stratus Dec 16, 2024, 5:38 PM

#

        try {
            client = new WebSocket("ws://localhost:8124");

            client.onopen = onOpen;
            client.onerror = onError;
            client.onmessage = onMessage;
        } catch (e) {
            trace(e);
        }

#

yep!

velvet lodge Dec 16, 2024, 5:39 PM

#

very weird - maybe i screwed something up when i moved the file over, not sure

toxic stratus Dec 16, 2024, 5:39 PM

#

velvet lodge Dec 16, 2024, 5:40 PM

#

well, good that wss works i guess, thats one less thing...

toxic stratus Dec 16, 2024, 5:42 PM

#

maybe, the whole situation seems odd to begin with 😄

velvet lodge Dec 16, 2024, 5:43 PM

#

what lib are you using for your server?

toxic stratus Dec 16, 2024, 5:44 PM

#

same lib

uncut phoenix Dec 16, 2024, 5:45 PM

#

Wait you are actually using core-haxe/websockets?

toxic stratus Dec 16, 2024, 5:45 PM

#

https://github.com/core-haxe/websockets/blob/main/src/ws/server/WebSocketServer.hx

toxic stratus Dec 16, 2024, 5:45 PM

#

uncut phoenix Wait you are actually using core-haxe/websockets?

yeah

uncut phoenix Dec 16, 2024, 5:45 PM

#

i am still using the one on ian's github

#

should i move lol?

velvet lodge Dec 16, 2024, 5:45 PM

#

oh, hang on so you are using core-haxe/ws for server too... right...

toxic stratus Dec 16, 2024, 5:45 PM

#

uncut phoenix should i move lol?

not terribly important i don't think

uncut phoenix Dec 16, 2024, 5:45 PM

#

tbf i would probably just use http if I knew better a few weeks ago

toxic stratus Dec 16, 2024, 5:46 PM

#

core-haxe one is still incomplete, i moved for a reason but i've been running it for so long i can't recall what it is now

velvet lodge Dec 16, 2024, 5:46 PM

#

toxic stratus <https://github.com/core-haxe/websockets/blob/main/src/ws/server/WebSocketServer...

hang on - there are no impls here...

toxic stratus Dec 16, 2024, 5:47 PM

#

https://github.com/core-haxe/websockets/tree/main/src/ws/server/impl/nodejs

#

😄

#

oh lmao

velvet lodge Dec 16, 2024, 5:47 PM

#

yeah

toxic stratus Dec 16, 2024, 5:47 PM

#

#

i may have done this myself then

velvet lodge Dec 16, 2024, 5:47 PM

#

ah, you are using the node server

#

still think there is something weird going on though... i still would have expected that to work with ws i think

#

https://github.com/core-haxe/websockets/blob/main/tests/cases/TestClientWithNativeNodeServer.hx some semi decent unit tests there thought 🥳

velvet lodge Dec 16, 2024, 5:50 PM

#

uncut phoenix tbf i would probably just use http if I knew better a few weeks ago

?

uncut phoenix Dec 16, 2024, 5:51 PM

#

Ah, for this specific project an http server would have made more sense compared to a websocket one

velvet lodge Dec 16, 2024, 5:51 PM

#

right

uncut phoenix Dec 16, 2024, 5:51 PM

#

it has just complicated things, nothing wrong with websockets for other stuff though

toxic stratus Dec 16, 2024, 5:53 PM

#

velvet lodge still think there is something weird going on though... i still would have expec...

I wonder if the server is booting in wss mode by default

velvet lodge Dec 16, 2024, 5:53 PM

#

not sure, maybe... thought i use it in unit tests and its ws ther

#

i just ran the unit tests for hxcpp and they pass

#

and they use ws

#

#

actually all those haxe targets UTs pass (which is pretty cool actually)

toxic stratus Dec 16, 2024, 6:00 PM

#

very odd

velvet lodge Dec 16, 2024, 6:05 PM

#

i wonder if git hxcpp has broken something?

toxic stratus Dec 17, 2024, 2:53 PM

#

velvet lodge

i'm pretty confused by the server here

#

how does the server unit test work when there's no implementation

#

i've shadow'd the module locally so that's why it works for me

#

oh no i haven't

#

the package structure is misleading x)

#

nodejs implementation is under ws/extern/nodejs

#

and your unit tests references the externs file instead of the impl file

#

https://github.com/core-haxe/websockets/pull/2/files

#

lol

#

i never saw that

#

@velvet lodge could you run the tests there, i'd like to see if those pass 👀

#

Oh! I understand why normal ws fails

#

I'm trying to connect to through the ssl server

#

that pr will have the same issue i suspect

velvet lodge Dec 17, 2024, 3:42 PM

#

toxic stratus how does the server unit test work when there's no implementation

it uses a prebuilt nodejs server, not its own one (which as you say doesnt exist). I want "two sets" of UTs - 1 set against an external, 3rd party lib - nodejs in this (to verify compatibility) and another set against its own server impl

toxic stratus Dec 17, 2024, 3:51 PM

#

yes, all makes sense now

#

and i've got ws and wss running locally

#

now i need to figure out why db isn't working

#

i should really test things out more often on native 😅

#

debugging stuff over here is a pita

velvet lodge Dec 21, 2024, 11:11 AM

#

Memory exhausted.
 try HXCPP_GC_BIG_BLOCKS.

this was a fun one

#Core Haxe