#archive-library-discussion

25085 messages · Page 7 of 26

oak quail

ah

yikes

ruby terrace

and there's no way to just freeze the user object because you'll get multiple GMU / PU per user change depending on how many servers they share, so it is only successfully freezed on the first update recieved, the rest already reference an updated user object

opaque copper

The tag property on User is returning null for a partial user? I understand it can be partial but discord docs never mark the username/discriminator property as nullable?

copper laurel

yes

Because if we had the full Discord structure it wouldnt be partial

patent halo

not sure if this is possible or its just "hard" to do, is it possible to manage the shards through PM2 .
Like if the library could implement it to auto shard and tell the PM2 to spawn N of processes

ornate topaz

what benefits would what have?

apart from having to write completely new way of having shards communicate between each other, which is supplied by the fact that main process spawns child_processes (or worker threads) which in turn allows communication directly in the processes, without any external abstraction

real jetty

In version 13, would you be able to give us more control over the caching? Some of the more popular bots that use discord.js use a crazy amount of memory, and most of the cache is useless because the bot doesn't use it. Even if it was just an object that said okay I want channels, roles, guilds etc etc

Then you could also ignore events where the bot doesn't want the cache for it

ruby terrace

I believe you can achieve that already (and will be more selective naturally in v13) via intents. If you don't need a message cache, don't add the messages intent. Speaking of which, that is something that needs to be heavily considered for v13. While many people will use the built-in INTENTS.NON_PRIVILEGED (IIRC that's what it is), granted not what should be done and most likely unneeded, it is totally possible, and recommend, to only specify the intents that you think your bot needs, only to find out djs breaks because it expected cache to be present and updated

copper laurel

Intents are one method of controlling caching as ckohen pointed out - if you never receive the events, your bot doesn't cache them. v13 is unlikely to change the caching structures and methods themselves though - this is something we're likely to explore more in the TypeScript rewrite

real jetty

Hmm okay thanks, I will look into intents and see if I can sort that out. Don't wanna have to buy a bigger server because my bot is using 200gb KEKW

(it's not actually)

copper laurel

You can also manually clear caches as needed

discord.js is built to expect things like guild/channels/roles to be in cache though

tacit crypt

The main things you can clear are users, members, maybe emojis, and presences. Those you can clear with little worry (except having to fetch members when needed)

real jetty

Ah alright thanks

quiet viper

Ghost Ping RIP

quiet viper
real jetty
quiet viper
unique axle

discord.js does currently not support swappable cache implementations or optional caches
this is a design goal to apply when we get around to rewrite the library.
that being said, you can always clear caches you do not need on an interval, as has been said above. not receiving data in the first place can also reduce cache size.

if you want to further discuss this on a level that's got nothing to do with discord.js development please use #archive-offtopic

wild flax
quiet viper
wild flax

Yeah now that makes sense lol

old hill

a native implementation of AbortController has been released in node.js 16
should this kind of thing be delayed until a future version of djs, since node 16 is so new?

and I guess also because swapping it out doesn't achieve anything of much significance aside from removing a dependency

warped crater

could polyfill it I guess

i.e check if it exists globally otherwise use the package

patent halo
old hill

so it was

I missed the

Some of the recently released features in Node.js 15, which will also be available in Node.js 16, include

question is still relevant though

raven juniper

Since master targets node 14 and I don't think there are any plans to make it target 15, this would be waiting until whatever comes next

ruby terrace

Seems like API v9 has been released to the clients (canary only for now). Since v13 isn't finalized yet, any thoughts on jumping directly to api v9 (obviously once documentation is present)?

sick merlin

the issue is at this rate, by the time all changes have been done to use v9, discord will release v10 and discord.js v13 will never come out
and theres no eta ( afaik ) for v9

vivid field

v9 has almost no breaking changes afaik, so switching over shouldn't take too long

copper laurel

v9 isn't likely to have THAT many changes

unique axle

let's start worrying about API versions when they're actually documented, thanks catThumbsUp

covert flint

Right, that's what I mean by "officially"

oak quail

on I didn't know there was another commit

there was one a few weeks ago

unique axle

We can start discussing implementations of threads once the features API is solidified and a PR on the original source repository discord/discord-api-docs, not some fork (which is already deleted)
A merge will of course require the implementation to be tested and the docs PR to be merged

empty viper

Node.js v16 is out also 👀

unique axle

that's also not relevant, because by far not LTS

empty viper

v16 will be LTS

unique axle

yes, at some point that might factor into library development. at the moment LTS is still 14.16.1

copper garden

Node v16 is scheduled to become LTS near the end of October 2021. D.js v13 should hopefully be released way before that happens

solemn knoll

i wondered how exactly a discord library like this on fetches "incoming" messages (like in this case with client.on("message", msg=>{...}))

copper garden

Why would you need to fetch a message?

solemn knoll

does it, like, fetch every latest message of each channel of a guild and compare it with the previous fetch

copper garden

No, it’s an event received via the ws

solemn knoll

ah makes more sense

ruby terrace

woah, sorry to spark such a controversy, thanks for the opinions (Souji, of course, was looking on a grand scale, although so far it seems like going to v9 might not involve a major change for the library somehow, all speculation for now though so I'll shut up)

oak quail

@ruby terrace interaction routes dont use the bot token anyway

WebhookClient is the easiest way to implement it afaik

otherwise youd have to copy all the methods

ruby terrace

That's an error in the latest PR's that add editMessage and deleteMessage, they should be passing auth: false to the request anyways

faint cipher

what's the point of resolve and resolveID if they depend on cache

ruby terrace

resolveID does not depend on cache, whereas resolve does as it is attempting to give you the cached instance

strange python

Hey. I was wondering if there is an official port of discord.js for deno. I found discordeno but I'm not sure if that is the official one

remote wasp

no, there isn't any

strange python

I see

Thanks :)

junior pumice

will v13 support slash cmds / interactions in general

unique axle
verbal solar
faint cipher

so resolve is useless

copper laurel

no it isnt

get must be passed an ID, resolve generically supports being passed any resolvable and returning the right thing

eg GuildMemberManager#resolve works with GuildMembers, Users, IDs

faint cipher

oh

empty viper

Hi, I'm talking about djs-next, what is tsyringe ?

raven juniper

?npm tsyringe

rich iglooBOT
vivid field
clever crypt

@vivid field resolved

void geode

Hey, do you know when stage channels will be supported by discord.js ?

unique axle

with the next major release (v13)
if you want to test it you can install the master (development) branch:
instructions and caveats: <#djs-help-v14 message>

worn bobcat

Why aren't the Discord embed types not sub stringed? Like this would make much more sense to substing the embed types not just trow an error saying that there are too many characters.

covert flint

Why should the library be responsible for something at the fault of the user?

wild flax

Its your responsibility to make sure the input is correct.

void geode
void geode

And, do you know if the parameter dynamic of ImageURLOptions will turned to true by default in V13?

remote wasp

no

raven juniper

There was a pr, or at least a discussion, about having it be true by default but it couldn't be done in stable because it's a breaking change

void geode

Yeah I understand, thank you

rough glacier

what was the reason behind this change?

vagrant holly

Encourage the user to handle errors properly, actually checking the message still exists before they delete it etc.

empty viper

You can change it to client.setTimeout(message.delete, 10000) no ?

slate nacelle

No

empty viper

why couldn't you 🤔

tacit crypt

Many reasons that are outside the scope of this channel

mighty sequoia

If i wanted to do this issue, what git commit message would I give it like feat or chore?

slate nacelle

feat

oak quail

so

threads

docs are available and looks like lib devs will have access to test them before full release (which will be in at least 2 months)

ruby terrace

Already working on adding them, might make a draft PR if someone doesn't beat me to it. I'm thinking to add threads into the channel cache since they are truly channels at heart, then have a #threads on text/news channels that is a ThreadManager. That way they can be accessed globally and in the guilds cache but still have references in their actual parent channels

ruby terrace

Also, definitely need to wait until after that disclaimer about potential breaking changes is removed

copper laurel

Yeah we get updates from Discord Developers too. They're API v9 though which although is supposed to have basically no breaking changes, isn't what v13 has been written to so far

Could be an easy switch, but I don't know if we'd want to wait another 2 months on deploying the other features

ruby terrace

So far (of course subject to change but hopefully not), I think switching to api v9 would be a non-breaking, non-major change for the library, its already using the PATCH method for suppress-embeds which is the only actually breaking change for v9.

But yeah I agree the release of v13 should not be delayed for it no matter what

mighty sequoia

Does a pr take 3 approvals to merge?

ornate topaz

there isn't any specific number for that

mighty sequoia
outer raven

When will package.lock be updated to lockfile version 2?

ornate topaz

probably not soon

since it's not really backwards compatible, and v16 isn't LTS yet

oblique prairie

Hey! When a client creates a shard, it emits messages to the debug event listener. One of them is IDENTIFY followed by the id of the shard over the total number of shards. However, this display method is confusing.
For example, let's say a shardManager creates 3 shards. The last IDENTIFY message sent is [WS => Shard 2] [IDENTIFY] Shard 2/3 which could mean that the 3rd shard wasn't created yet.
I think it should be changed to [WS => Shard {ID}] [IDENTIFY] Shard #{ID} ({ID+1}/{total}) where ID is the id of the created shard and total the total amount of shards to be created.

tacit crypt

That's not happening, as shard IDs are 0-based

Think of shards like an array, the ID being the index, the count being the length

oblique prairie

But 0/3 being the first and 2/3 being the last is really confusing and doesn't make any sense even if it's 0-based.

tacit crypt

Could we? Yea! Should we? ehh..
The docs also show it as 0-based, and it'd stay consistent with that and how the identify payload works. We could add a (normalized ${shardID + 1}/${shardCount})

tacit crypt
oblique prairie

The id is a 0-based cardinal number, you're right ;
in (position/total), position is an ordinal number tho, which means by definition that it has to be between the minimum value of total and total (included).The minimum value of total cannot be 0 so position must be included in [1 ; total]. Showing (value1/value2) implies that the two values are comparable and thus have the same meaning. An ordinal and a cardinal do not have the same meaning at all... Anyway, this debug line is just a detail.

flint zephyr

Hey guys I am making my own client and am here for some guidance (don't worry it's totally unrelated and not a discord.js competitor)
So the thing is, I am confused what to do.
My initial plan was to connect to the gateway and cache the session id. Whenever the bot connects after that, use Gateway Resume instead of making a new session.
But would this be a good idea? What does discord.js do?

unique axle

this channel is for the development of discord.js, not to help develop other libraries for the discord bot API

flint zephyr

Well then is there any other channel to ask such a question? Or is it banned D:

wild flax

Making your own client is already not really ToS conform, so it will be kinda hard finding a good place to ask such a question

flint zephyr

Why is it not TOS compliant? I'm contributing to open source and helping people.

wild flax

If you mean your own library instead, the discord API server is always a place to ask in #resources

flint zephyr

Oh yeah I mean library, sorry for the confusion

Thanks I'll take a look

worn bobcat

Why aren't the Discord embed types not sub stringed? Like this would make much more sense to substing the embed types not just trow an error saying that there are too many characters.

raven juniper

It's not the library's responsibility to check your input. It also makes it harder to maintain, if discord ever changes their limits

It's better in that way to let the API throw the error like it does

unique axle

to add to that there are a ton of different limits that need to be enforced. for example the "all characters in embed values can not exceed 6k"
in which of the up to 25 field bodies to you substring? why? do you just delete fields if you don't have characters left?
how do you handle the description? does it have priority over fields?
the title seems important, so maybe always use that... but what if field titles are already taking up too much space with the description, do you just show field titles? or field titles and values until that's depleted?
do you prioritise the footer over fields? or not?
etc.

drifting thunder

if its possible, function for that would be useful

empty viper

No it isn't

real jetty

Add event : on_bot_add

Or something linke this

stable escarp

what is that supposed to be

real jetty

That if someone adds a bot is happening an event

Sory translator mistranslated xD

lofty birch
stable escarp

you can listen to guildMemberAdd and filter bots? I don't think a whole new event is entirely necessary

undone ravine

That would already be very achievable by listening to the guildMemberAdd event and checking to see whether the user is a bot

undone ravine
stable escarp

plus I'm sure there's a way to make your own event. Come to think of it, I should do that

real jetty

But I mean that as someone will invite __ bot with this event __ something happens

And not that as the bot is already on serwer

real jetty

That's what sweat

stable escarp

so you want to know who added the bot? I'm not sure the API handles that, correct me if I'm wrong

raven juniper

You need to fetch audit logs

stable escarp

yeah

drifting thunder

this is 2 years ago, something may have changed

covert flint

If that were the case then I'm pretty sure they would've said something by now

outer raven
ornate topaz

I know

And that's why i said what i said

outer raven

but v14 is LTS

ornate topaz

Also no, node 14 does not come with npm 7

15 does

outer raven

but it prompts you to update when you run npm commands

and v7 is not in beta, it's the latest release

it's what gets installed when you run npm i npm -g

ornate topaz

Prompts you because node 14 comes with not the latest npm 6

And yes, correct, it's latest release, but it's different major, and as expected, has breaking changes

Which in this case rule out backwards compatibility

outer raven

does package-lock change backwards compatibility all that much?

This change only adds things it doesn't remove anything

and I believe if you use an older npm version it converts to version 1

ornate topaz

It doesn't convert

It warns about unknown version

outer raven

it did when I was using it

it automatically changed the file and warned about the wrong version

ornate topaz

Yeah ok. Still doesn't change the fact that lock file in such case would be constantly swinging around.
Npm 7 does not come with latest LTS, and we don't require node 16

outer raven

is node 16 even a thing?

copper garden
ornate topaz
outer raven

OH I thought the "Current" version was v15, was not aware of the update flush

ornate topaz

Both 16 and 15 are current rn, look at linked time table

empty viper

V16 doesn't really add features anyway

strange moat

It does add features, though their usefulness is questionable

wild flax

It’s a new LTS regardless. It’s not about features but the long-term support it has for us in case there is bugs or whatever

Since they will be fixed in an LTS compared to v15 for example

weak parrot

Region changer for VCs?
like
message.member.voice.setRegion("EU") ?

brisk pawn
weak parrot

oh

outer raven

there are multiple instances across the library where optional parameters are marked like this which does not work properly in the docs website (the name of the variable should be wrapped in []). Should I make a PR to fix all of these or is this somehow intentional?

remote wasp

You're confusing it with optional param, the ? here shows that it can be null

slate nacelle

No, it's using node-fetch.

outer raven
delicate agate 
New slash commands for bots! Performing actions with bots is easier than ever. Bot developers—when they’re ready, please don’t scare them—can have a list of commands available for their bots accessible by the forwardslash key (/). You won’t have to remember bot commands anymore and get laughed at when you get the command wrong :( !help…..!help….please

Any idea when and if this going to be in Discordjs?

loud jayBOT
tender field

it is this PR iirc ^

yep it is it (cf the pinned messages)

brisk pawn

When will djs update to v9 for threads?

Like how long after it comes out?

Hopefully it won't be like slash commands, it's been like 6 months since it released and the pr still isn't merged

warped crater

as always, when it's done

copper laurel

The initial announcement of the developer beta wasn't even 6 months ago, let alone actual release (which has still had more changes since). It also went through a complete rewrite via a second new PR

topaz coyote
worn bobcat

Why can't @everyone mentions be disabled by default? It would make much more sense to make a option to enable it

ornate topaz

this isn't channel for suggestions for discord

wild flax
strange igloo

Bringing the discussion here so it can get answered quicker than on GitHub.
As of now, <BitField>.has(undefined) returns true which may be problematic if you've made a typo in your code (e.g.: <Permissions>.has(<Permissions>.FLAGS.MANAGE_SERVER)).
<BitField>.constructor.resolve returns defaultBit if typeof bit === undefined.

Now, how should the library behave if undefined or [..., undefined, ...] are given to BitField#has. My first idea was to throw an error, like we did for earlier versions with string flags, but this message from Monbrey > #djs-help-v14 message kinda confused me.

  • Should <BitField>.constructor.resolve throw an error when typeof bit === undefined ?
  • Should we silently ignore undefined bit(s) and do not count them ? (<BitField>.has(undefined) === false)
brisk pawn
worn bobcat
vagrant holly
strange igloo
lofty birch

Hi. I've noticed more than once that my server IP become temporarily banned from discord due to exceeding ratelimits, specifically due to multiple webhook clients (which are dynamically generated by user configured webhooks). I would assume this is because of a global limit of 50/sec? Does the library not handle this?

ornate topaz

don't think that lib tracks that

lofty birch

it probably should given that I'm using it's WebhookClient class

warped crater

Unfort. its hard to track but global rate limits are handled

ruby terrace

Each WebhookClient uses its own #api internally, which has separate ratelimit handling

but global limits should be taken into account IIRC

warped crater

which means you'll hit between 1 and N 429s for global

but N tends to be fairly small...

you def shouldnt be running into this frankly

cc @tacit crypt

tacit crypt

You got... 429d... on your bot token, by using a webhook client...?

warped crater

they got the 10k ip ban

tacit crypt

Thats on a bot token tho?

warped crater

no, its ip

lofty birch

it's an IP ban, yes

tacit crypt

Is it? damn

Are you sharded?

lofty birch

only 6 shards

tacit crypt

And you use the same webhook on all 6 shards?

ruby terrace

well 4711 was only merged to master, so djs 12 won't have that fix yet

warped crater

oh, right

lofty birch

ah so it's not in stable, that explains it

might have to merge that specific branch into a fork of V12 because it's very inconvenient

lofty birch
oak quail

theres a new clientoption to handle the 50/s global rl

oh i think thats what ckohen mentioned

but yeah that probably wont help u

are you getting the "you have been temporarily banned from our api for consistently exceeding our ratelimits" error?

lofty birch

yes

that PR would at least back off

the solution might just have to be making some queue system that runs on a cheap VPS that handles these across all shards, as those obviously make this a lot harder to handle

oak quail

is there a reason u arent just sending via the bot

would it really hit 50/s?

lofty birch

I mean it has

when night helped me debug it the first time it happened, almost every one of the ratelimit requests occurred on webhook routes

oak quail

hmm

do webhooks have ratelimit headers now

lofty birch
oak quail

u can still send cross server via the bot tho? 🤔

lofty birch

the bot isn't guaranteed to be there

oak quail

ah

so the server provides the webhook url?

lofty birch

yes

ornate topaz

cough didn't that go slightly offtopic?

lofty birch

you're right, was just asking here to find out if it was a lib thing and it turns out there was a PR for master, just obviously not on V12 yet lol. thanks for the input

spiral oak

can MessageType be updated to include all the new ones?

wild flax

You need to be a bit more specific than that

spiral oak

These are the current djs ones

But the discord doc has a load more now, including ones like reply which would be very useful

slate nacelle

Those will come with v13, for example reply and application command are already there.
The thread types will come with threads, which will still take some time and no idea what the last one is.

spiral oak

👍

tacit crypt

INVITE_REMINDER is related to a system channel flag

remote wasp

The InteractionReplyOptions links to WebhookMessageOptions in docs, which isn't completely correct as the latter one has some extra fields. Vaporox and I had a talk about it (#djs-help-v14 message) and the problem was that making a new typdef would result in code duplication for the third time.
One potential solution that we arrived at was creating a new base typedef that will have all the common fields which MessageOptions, WebhookMessageOptions, InteractionReplyOptions, and ReplyMessageOptions (yet to be merged) can extend. This will result in no code duplication and a correct documentation.
I wanted to discuss about it first before working on it just to make sure that this is what we want.

outer raven

Don't really know where to put this but is it possible to improve the embed sent with discord.js.org links so that it shows more information about that link?

wild flax

Not really no. We can't really do SSR and pre-generating all the pages depending on stable/master/version/branch doesn't seem necessarily worth it.

It would also make the embed unnecessarily big, any reason why this would be particularly useful?

outer raven
wild flax

Yeah but there isn't necessarily a big benefit in this

outer raven

It would just look nice and would help people quickly understand what they're gonna find once they click the link

wild flax

The benefit doesn't really outweigh the work required for it here sadly

outer raven

Fair enough, I'm not sure how these docs are even generated so it might be harder than it seems, thanks for the feedback!

remote wasp
outer night

it's been 12 hours, give it at least another day before bumping

pale seal
remote wasp
pale seal

I'm having the issue too. Seems to be an issue that comes back all the time xD

outer raven

Quick question: what's happening in the options.types type in the ApplicationCommandOptionData interface? It has 2 values and 1 of them seems to link to the other and, when writing in TS, we don't get autocomplete like we do with permissions, intents, etc

vivid field

ApplicationCommandOptionType are the strings that you can pass, ApplicationCommandOptionTypes their corresponding numbers

outer raven

ah gotcha, why aren't they autocompleting then?

vivid field

they should be

outer raven

Could you try to reproduce this? Maybe I'm doing something wrong

vivid field

oh, could actually replicate

but i guess that's a problem with typescript then, not with discord.js

outer raven

Well I can't get to the value of ApplicationCommandOptionTypes since it's an import and for some reason my IDE doesn't let me open discord-api-types/v8 but maybe there's something wrong with that?

You could try declaring the types directly in index.d.ts instead of importing, not sure if that would fix it tho

vivid field

you can open it, just ctrl+click the import

outer raven

yeah I know but it doesn't work

just doesn't do anything

vivid field

but it shouldnt really matter where the enum is declared

outer raven

I can't find a folder called discord-api-types in my node-modules either

vivid field

even tho it will have to be declared locally anyway with discord-api-types v0.19

then you didn't install it probably

outer raven

I think I did tho

lemme try again

yeah just reinstalled and the import still has any value 🤔

I can ctrl click all imports except for discord-api-types/v8 and ws

I think the package isn't declared as a dependency

ruby terrace

I was having some weird issues with ts imports too, both in master and -next (it is a dependency and is in the modules folder though)

outer raven

Is it a dependency? I can't find it in my package-lock.json

ruby terrace

dev

vivid field

it's a dev dependency

outer raven

well those don't show up, that might be why

They're not getting imported when installing the package

vivid field

yes, you have to install discord-api-types and @types/ws separately if you don't want them to fall back to any

outer raven

ooooohhhh

it works now

it was that simple huh

lone vector

I'm curious on the decision for how Role#edit() is made to work, since from what I can tell reading the code for it, it always sends the permissions field to the API although the API considers it optional, so just trying to edit a role's colour or name can cause a Missing Permissions error if the bot doesn't have at least the exact same role permissions as the role it's attempting to edit, or maybe I'm just missing something/doing something wrong to cause this?

solemn oyster

Continuing from #archive-offtopic message

@old kelp proposed merging create and edit into upsert, as it's the term for "create or edit".

However, since the edit part isn't a patch, but rather a full rewrite (override, if you will), it matches set's description, which also creates.

This also wouldn't break consistency with existing methods in the library, e.g. member.roles.set. While I don't recall upsert being used anywhere else.

What are your thoughts? @tacit crypt

tacit crypt
old kelp

so are we going with upsert?
cc @solemn oyster @tacit crypt

tacit crypt

thing is, with permission overwrites, create and edit do the same

unique axle

no BOYEbomb

tacit crypt
unique axle

no

tacit crypt

Souji dear

unique axle

vlad dear, i broke that discussion loose 2 times, and every time i in the end saw that space was right and i was wrong, they are not the same

tacit crypt

Look at the source code for the overwrite methods, they do 1:1 the same code??

old kelp
tacit crypt

Casein point

old kelp

bump again kyra, vlad, so are we going with upsert now or..

tacit crypt

iirc i said upsert too at some point, I can't speak for kyra or crawl tho

old kelp

@unique axle could you elaborate on your point?
is it not same just for making it more verbose to users, because they internally work the same, otherwise i'm missing something?

unique axle

nvm, that was GuildChannel#createOverwrite and GuildChannel#updateOverwrite (they don't do the same, hence why they still both exist)
which:

thing is, with permission overwrites, create and edit do the same
sounded like potatodetective

tacit crypt

Souji dear..

unique axle

no.

tacit crypt

WDYM NO??

Look at the code

unique axle

i literally argued about that for hours with space and i lost

they don't do the same

update retains other bits but the ones specified, create replaces them

tacit crypt

Wat

OH, I see

Still, code deduplication would be ideal

unique axle
tacit crypt

Yeah the difference is that update appends it, as space said, while create just..creates

But the end API calls and checks are the same in a way

So, those parts should be deduplicated in a private function

ashen garden

add an event "raw gateway response" or something

basically a way to listen for everything the api sends back to us

ah, is it undocumented?

ye exactly

thanks

old kelp
tacit crypt

No, edit and create do different things, but their end code is the same except for one line

old kelp

so should edit call create then?

tacit crypt

you should have upsert as the base func, create call it, and edit call it with the existing overwrite

Let me get you example code

Think ```js
class Example {
upsert(userOrRole, permissions, reason, old = null) {
userOrRole = this.guild.roles.resolve(userOrRole) || this.client.users.resolve(userOrRole);
if (!userOrRole) return Promise.reject(new TypeError('INVALID_TYPE', 'parameter', 'User nor a Role'));

    const type = userOrRole instanceof Role ? 'role' : 'member';
    const { allow, deny } = PermissionOverwrites.resolveOverwriteOptions(options, old);

    return this.client.api
        .channels(this.id)
        .permissions[userOrRole.id].put({
            data: { id: userOrRole.id, type, allow: allow.bitfield, deny: deny.bitfield },
            reason,
        })
        .then(() => this);
}

}

class PermissionOverwrites {
update(options, reason) {
return Example.upsert(this.id, options, reason, this);
}
}``` but you'll need to make it prettier (theres a PR that makes it to where we can specify the type, so.. that might be good too)

I think this would be accurate-ish

could be cleaner, shoving everything in an object too

CC @wild flax @solemn oyster

old kelp

shouldn't the base method be in PermissionOverwriteManager and make that call in PermissionOverwrites though?

oh, that's what you technically did.. i'll come up with a cleaner implementation and re-ask for review, thanks 👍

tacit crypt

Yeah, think Example as the manager

outer raven

was the code link in the docs removed?

I can't seem to find it for some reason

oak quail

also in the "Links" section of the welcome page

outer raven
oak quail

this?

never knew it existed tbh, thats useful

outer raven

ye I can't see that

lmao really

but I don't see that button for some reason

oak quail

try hard reloading?

outer raven

yep did exactly that and it worked

thats odd

ty

unique axle

resize your window, something weird happens there sometimes that pushes it farther out

outer raven

it was maximized ¯_(ツ)_/¯

grave tree

What discord gateway link is using discord.js?

wild flax

gateway link??

grave tree
wild flax gateway link??

when you are connecting to the gateway and receiving payload with the websocket, you use a link like wss://gateway.discord.gg/?v=8&encoding=json

wild flax

Yes, well that very much depends

Encoding depends on if you have zlibsync installed, but overall we are on gateway version 8

grave tree

oh thank you

warped crater
wild flax

lol, the 2nd is not so relevant

You'd be better off just looking in the Constants or WebSocketShard

warped crater

yeptune fair

wild flax

Convoluted mess

NiimiNotDead

warped crater

one day

grave tree

is discord.js master branch updated to gateway v9?

keen cosmos

there's a PR for it

loud jayBOT
solemn oyster

There are some backwards incompatible changes with gw v9 that will be released over time, and stable doesn't even use v8 yet

keen cosmos

oh, gateway, not api, my b

silk wren

How does D.js get the GuildMember#joinedAt ?
Of course it has to be exposed by the API itself, but I'd like to hear where x)

unique axle

joined_at ISO8601 timestamp when the user joined the guild

grave tree
next inlet

@solemn oyster mind jumping in a VC with me for a sec?

uh,,

solemn oyster

Is this going to be related with the library?

next inlet

yes

give me a sec, gonna restart discord

solemn oyster

Ok

solemn oyster

We concluded a structure called ApplicationCommandOption (as opposed to a plain object with a typedef) would be a sensible way to approach this

And it'd also refactor away ApplicationCommand.transformOption (they'd be transformed inside ApplicationCommandOption's constructor) as well as allow us to add isEqual in them for fast deep comparison

The idea behind checking for equals recursively is so we can dynamically build a framework which updates the registered application commands on Discord automatically, without us needing to write migrations or pushing all commands in every log in, which can be rather inefficient

copper laurel

So it would load files like a regular framework, then fetch API and compare?

next inlet

yes that is pretty much the thought

vivid field

Sounds reasonable for received options, but how would that approach handle options passed by the user?

solemn oyster

Isn't that pretty much framework independent? It's not like regular users can update one bot's slash commands

vivid field

I mean the users of the library

You shouldnt need a framework to create/update application commands

copper laurel

Need, no, but you never needed one for regular commads either

It just makes it easier

vivid field

That's why I'm asking, because from what I understood, for commands.edit()/.create()/.set() you would still need a transformOption method or similar

sly forge

The ApplicationCommandOption structure sounds great, but handling received interactions should be left out for the user to implement.

ruby terrace

I assume you would then need to pass it an instantiated ApplicationCommandOption which would have a toJSON method for readying it for discord. (not sure if I'm a fan of this)

ruby terrace
vivid field

I guess that would work, but constructing a new class instance only to convert it back to an object right after also sounds a bit expensive

ruby terrace

yeah, that's why I'm not exactly a fan

solemn oyster
solemn oyster
tawdry pasture

Will the BLURPLE string in
ColorResolvable be updated to match the new blurple?

wild flax

Yes

Mind you, this will most likely happen in a non-backwards-compatible fashion since we are increasing the major version

tawdry pasture

So Ill have to switch my v12 bots to using the new color manually then?

wild flax

No. It means we’ll switch the color and when you update you automatically have it

Usually you aren’t allowed to do that as a library, since users might still want the old color without us forcing it on them

unique axle

will have to do some figuring out regarding the new default icons, they're not numbers anymore nkoBruh

eh, v13 is a major anyways, might as well flip colour codes and avatars

tawdry pasture

Ah got it. thanks!

oak quail

the client has always used /assets instead of the cdn ones

outer raven

I noticed the default avatar URL no longer works. Is this a bug with the library or discord?

By "no longer works" I mean it links to the old one(s)

ornate topaz

you could read messages just above yours, and judge on your own why it links to old ones

outer raven

oopsie, didn't notice, sorry

ornate topaz

yes, that is /assets

outer raven

Adding the bigint type for RoleManagers made things slightly more annoying for typescript devs, is it possible to remove this type for this and other managers that have this?

vernal atlas

what

ornate topaz

what are you talking about

outer raven

Whenever I try to access a guild member's roles, the manager has those 2 types which makes it so TS always throws an error when trying to remove/add roles for example

Talking about the master branch btw

wild flax

Please post the associated code with it

outer raven

ok sorry I missed some context. I'm getting this when accessing a guildmember through an interaction, for which the roles object is a Snowflake type sometimes, which comes with that `${bigint}` type

the line is just this

wild flax

Yes.

outer raven

Is this intentional?

ornate topaz

fairly sure it is, and it's array of snowflakes when there is no bot

outer raven

How would there be no bot?

ornate topaz

interactions don't need a bot

wild flax

Interactions don't require a bot connected to the gateway or in the guild

outer raven

even if it's a commandinteraction?

wild flax

of course

ornate topaz

what difference does that even make here

outer raven

that we have to specify the type for the rolemanager every time we want to add or remove roles which is slightly annoying

wild flax

well, it is accurate thats for sure

your code will break if you don't check first

outer raven

well in my case my bot is only in 1 guild and I'm sure it will always be there and have access to interactions so I don't think any checks would be needed

wild flax

Sure, whatever, but that doesn't mean we can change the typings just because you have no use for it

outer raven

Yeah of course but it seems like such an edge case that will cause so much trouble for other users and isn't even documented properly on your docs website

ornate topaz
outer raven

Yeah exactly, doesn't specify any properties for that "Object" so it's hard to know

wild flax

How is it an edge case? If someone invites your commands via the oauth scope without inviting your bot, you need to check.

outer raven

You can do that?

wild flax

Yes.

outer raven

monkaS

ornate topaz

but it's an Object. it does not have any methods GuildMember would

outer raven

it could be the APIInteractionGuildMember object which is what is given in TS

untold python

I dont know if this fits here, but is there a version of discord.js that doesnt use the fs module? I need to run it in the browser, which doesnt offer the fs module

frank turret

webpacked versions of d.js aren't supported anymore and don't work

untold python

aw

ill write my own wrapper then, thx anyways

remote wasp

@untold python I don't think you can make any request to the discord api in browser now

remote wasp

yeah

cloud nest

Would it be thinkable to add an APIEvent enum containing all currently available events? That would simplify event handling implementations. Possible benefits of such an implementation (and the issue I am currently facing) are described at https://github.com/microsoft/TypeScript/issues/10570#issuecomment-769442192

I do see that solely supporting such an enum in the library itself would be a breaking change. However, one could merely provide support for using the enum for listener registrations as an alternative way to the current one.

Admittedly, this is a bit of a minor inconvenience, but it bothers me enough to propose such a feature.
I am quite new to d.js and apologize if such discussions have already taken place.

Also, I'd be happy to create a GitHub issue if need be. Thanks!

slate nacelle
empty viper

:/

cloud nest
slate nacelle Would `keyof ClientEvents` not be sufficient here?

I initially thought so too, but

abstract class Parent {
  abstract event: keyof ClientEvents;
}

class Child extends Parent {
  event = "message";
}

throws an error. TS cannot infer that event in the Child class is indeed a keyof ClientEvents, but interprets it as a string instead.

Instead, you need to assert the type manually each time:

class BetterChild extends Parent {
  event: keyof ClientEvents = "message";
}

It works, but it is an inconvenience that could be avoided by using an enum that would allow for TS to automatically infer the correct type

slate nacelle

Using event = "message" as const; seems to work.
This seems more like some "TS is not smart enough to infer that properly" than something we can do.

cloud nest

That's totally true, see the issue I referred to above. The enum is only a workaround to TS being incapable of figuring out the obvious on its own

outer raven

Would it be possible to create a latest pre-release with the latest commit from master and publish that to NPM so that we can more easily use that version? The website I'm currently hosting my bot on doesn't seem to support installing repositories as dependencies

remote wasp

That's not possible but you can fork the repo and publish your own fork of djs on npm under private scope

outer raven

How is it not possible? I've seen multiple packages do this

remote wasp

I don't know about that but publishing unstable changes doesn't sound like a good idea and it will also mess the versioning

outer raven

You can have a latest release that always gets updated with every commit to master and label it as a pre-release. This wouldn't affect the stable version and thus the version that would be installed through regular npm i discord.js

ornate topaz

which packages do that?
do they have breaking changes between commits that are published under that pre-release?

outer raven

Unfortunately I can't find one anymore but I remember seeing a pretty common package with a latest release on github and npm

outer raven
wild flax

Thought about it before already, decided against it

outer raven

Any reason why?

wild flax

Yeah, if we ever have some sort of hotfixes because our master isnt stable, we need to think about re-releasing it constantly

It's not worth it, you are much safer just pinning a commit that works

outer raven
wild flax

Yeah thats not really desirable for us

We don't want a new release on every commit

outer raven

Fair enough :/

real pilot
copper laurel

It's by design. Slash Commands are designed to be usable with a bot - headless functions running on CloudFlare workers for example

ornate topaz

*without

copper laurel

yeah ty, without a bot

real pilot

That makes sense, I just put a message in the interactions channel too, but is there a way to handle when someone adds your commands to a server even if they don't authorize the bot scope? Like a webhook, etc.? If so, can the library handle it and emit an event?

copper laurel

No, not in discord.js

ornate topaz

don't think there is a way to get an event for when someone adds an interaction without a bot

copper laurel

...unless it emits the ws event anyway?

ornate topaz

what ws event

copper laurel

well any of them

ornate topaz

i mean, there is intehration create, not sure if it's relevant

copper laurel

INTERACTION_CREATE, APPLICATION_COMMAND_CREATE etc

ornate topaz

but then again, if there is no bot, there is no ws

copper laurel

Yeah that I wasnt sure on, makes sense

ornate topaz

In many cases, you may still need a bot user. If you need to receive gateway events, or need to interact with other parts of our API (like fetching a guild, or a channel, or updating permissions on a user), those actions are all still tied to having a bot token. However, if you don't need any of those things, you never have to add a bot user to your application at all.
https://discord.com/developers/docs/interactions/slash-commands

wild flax

Thats not true

the WS event emits either way

ornate topaz

well, but if you don't have the bot app, you don't get the event (unless you connect to ws on your own, obviously)

and do you get the ws event from guilds you aren't on?

real pilot

Got it, makes sense. My app does still have a bot user since it needs it for some functions, but it also has commands that work perfectly fine without the bot user. What I'll just do is check each interaction, and if it needs a bot user it'll reply saying it needs additional authorization and send the oauth link for a server admin to fix it. Thanks guys!

lofty birch
copper laurel

@oak quail moving discussion
TextBasedChannel handles it like this

apiMessage = APIMessage.create(this, content, options).resolveData();
if (Array.isArray(apiMessage.data.content)) {
  return Promise.all(apiMessage.split().map(this.send.bind(this)));
}

So I guess reply with split would have to be

if (Array.isArray(apiMessage.data.content)) {
  const split = apiMessage.split();
  return Promise.all([this.reply(split[0]), ...split.slice(1).map(this.followUp.bind(this))]);
}```

Unless you can just send everything as a followup?

ruby terrace

You have to send a reply first (or edit the deferred reply)

copper laurel

I think followUp edits a deferred one, but if it can't send an initial it'd be this

If it's an editReply with split data they can all be followUps

...no nvm

ruby terrace

edits shouldn't take split options IIRC

followUp did edit a deferred one in the past, but I think that was for backwards compatibility, I think it has changed

copper laurel

So use case - deferring, then sending a split message

Can that be handled and if so how?

ruby terrace

It could be handled, but you'd run up on weird situations where there are messages in between

followUp should honestly be calling .webhook.send with the created apiMessage after data and files have been resolved, and Webhook#send line 172-174 should move outside the else statement

kindred lagoon

I really have a very huge bot and it is known as well and there is play sound when it starts the result is a very large pressure on the server so if you have a solution I will not reprogram it in c ++ language I am sure that you have a problem playing sounds so always the huge bots do not use your library

strange igloo
kindred lagoon I really have a very huge bot and it is known as well and there is play sound wh...

discord.js voice is not designed to be the most powerful broadcast solution
If your needs are above the expected target, feel free to design your own solution or use an existing one (Lavalink for the most popular one)
There are no flaws in how discord.js handles voice features
Also, sending encrypted audio data will always require some resources (in CPU for the most) no matter if it's being made by discord.js or any other library/solution, so the issue of the pressure will stay as long as you don't split your bot logic and your audio between two machines, or if you don't upgrade your current machine resources.

kindred lagoon
real jetty

Would it be possible to make an auditLogEntry event for Client?

warped crater

No.

the client only has events for websocket dispatch packets that come in from Discord

there is nothing to signal audit log creation

real jetty

alright, i'll try to manually make it then

copper laurel
ruby terrace

you can, if you resolve the data and files in followUp, it will use that resolved data and files in Webhook send, it doesn't re-resolve

copper laurel

Hmmm fair enough, I'll take another look.

real jetty

Not sure if this has been brought up before, but something that would be really useful are hooks. Here is a use case:

When discord.js gets the users, you would have a $beforeUserLoad(UserManager) hook where you can inject your own data into users. Then if you have sharding, it would shard all the users with your injected data
It would be the same for guild, so $beforeGuildLoad(GuildManager)

ruby terrace

Extendable structures are discord.js's way of doing exactly that

real jetty

You can extend the structures but you cant add the data before it gets sharded afaik

copper garden

Wdym when it "gets sharded". When you shard, all events and results of api calls are sent directly to the corresponding shard process

real jetty

Essentially when all the discord users are cached, i would want to have a hook where i can make a request to my own api and map that data to users

copper garden

The only issue I see with that is that it’s async. Constructing/caching a user wasn’t made to perform async calls

Can’t you just fetch the data on demand?

real jetty

yeah i could shrug

quiet viper

I have a suggestion, regarding too the broadcastEval in the Sharding Manager, you should add a optional option, where the promise should be resolved/rejected after a certain time, bc when this does not happen --> we have to many event listeners and this could cause a memory leak

ornate topaz

imo such timeout should be there by default

quiet viper
ornate topaz imo such timeout should be there by default

no there isnt, I can reproduce the error, go on worker mode, use a code which causes a dom expect error, broadcastEval(this.guilds.cache) --> error, no response.
Do this 10 times and voila, most of the machines will give a event listener error or crash

ornate topaz

what that has to do with my answer

i said "should" not "is"

in response to you suggesting there being one only if you pass the timeout time as the parameter

locking you out for no real reason if you forget to pass it

maiden loom

Hi, how can I download the v13 version ?

ornate topaz

there is no v13 version yet. you can however install the version from master branch, more info in #djs-help-v14

maiden loom
ornate topaz
maiden loom
ornate topaz
maiden loom
quiet viper
ornate topaz

i didn't say anything as to if it will be there or not. it was a suggestion

quiet viper

okay, sorry for the misunderstanding

clever spoke

it already does tho.... but just on master so i dunno if that counts

manic thistle

the typings for PermissionOverwriteOptions seem to not match the actual js implementation

wait whoops i forgot to put the typings, sorry

there's also this function which uses a completely different type in the options parameter?

remote wasp

that PermissionOverwriteOptions interface is defined in typings but isn't used anywhere. Whereas the typedef PermissionOverwriteOptions (plural) is defined as PermissionOverwriteOption(singular) in typings. weird 🤔

manic thistle

OH

actually nevermind, i didn't realize the definition was created in the same commit so it couldn't have been the parameters changing or something
it seems to have been a mistake that was added when the function was created and nobody really noticed lol

remote wasp

I looked through all of it and I think the current interface for PermissionOverwriteOptions (plural) should be removed from the typings (I searched it up, this interface isn't being used anywhere in the codebase) and the interface PermissionOverwriteOption (singular) should be renamed to PermissionOverwriteOptions so that it matches its jsdoc typedef. Just want to make sure that I'm not missing anything here. 🧐

manic thistle
remote wasp

that's fine but it doesn't solve the issue above

manic thistle

my bad
yeah, it seems like the interface PermissionOverwriteOptions was only used as a parameter for GuildChannel#overwritePermissions anyway, which was replaced by OverwriteResolvable (More specifically it was replaced by the interface OverwriteData, as it has the exact same properties, minus the new type property and the fact that allow and deny are now optional) so there should be no problem with it being removed, with PermissionOverwriteOption taking its place to match the documentation

ruby terrace
slate nacelle

Sounds good to me

oak quail

since members might not have joined_at now, should this be changed?

i suppose they still arent "full members" but it is a member object

manic thistle

I was inspecting the typings when I noticed the function Permissions#serialize returned the type Record<PermissionString,boolean> and that there was an interface PermissionObject (which wasn't used anywhere in the code) with the exact same type .
After inspecting the history of the function and interface, I found out that it was in fact previously used as the return type, but was removed (along with the function) when the Permissions class extended BitField<PermissionString> and later added back with this type.
Since these are the only two instances of Record<PermissionString,boolean>, should PermissionObject be removed?

strange igloo
manic thistle
real jetty

What is the point of removing timeout?

raven juniper

Keeps it in line with other methods, also to prevent some problems that arose (eg if it was deleted before the timeout)

strange igloo
wild flax

No

The output isn't stable

And may change in the future depending on our needs

strange igloo

Oh okay

wild flax

If you want to parse an emoji, we do expose the regex itself on the class

to at least get the id

outer raven

@copper laurel do you know what the new buttons will look like? Asking bc of your draft pr

copper laurel

Yes

They look like other Discord branded buttons. Can be blurple, green, red or grey

remote wasp

Found this while working on the nsfw_level PR. Same issue there. When I move the property assignment inside #_patch it logs DEFAULT correctly otherwise it's just undefined

slate nacelle

It should be in _patch. Must have slipped through jynEHEH
Discord only sends unavailable guilds in ready, which we then patch with their relevant guild creates.
So only newly joined guilds would have that property set if it's in the constructor.

remote wasp

I see, I'll update the PR. Thanks 👍

real pilot
copper laurel

Not that I'm supposed to share

Sorry! Should be available soon

real pilot

Ah okay, I didn't know it was private info still haha

outer raven

They’ll probably be “released” on Tuesday

Why does CommandInteraction#channel have a Channel type instead of a text based channel type (TextChannel, NewsChannel or DMChannel, maybe ThreadChannel idk)?

copper laurel

Because it's a getter for this.client.channels.cache.get(this.channelID) which is a Collection<Snowflake, Channel>

And its inherited from Interaction

fringe star

Where can I post bugs?

solemn oyster

What bugs?

fringe star

.permissions doesn't work correctly

For some members

It returns bitfield 0

Even if they have certain permissions

I really need a statement on this, please

slate nacelle
fringe star

Ohk

fringe star

Can I send a reproducible example here

slate nacelle

sure

fringe star

I have almost all permissions still the bitfield is 0

This thing worries me when I check for permissions in certain commands

All checks return false

bruh moment

remote wasp

can you do npm list discord.js and send the hash here

fringe star

Ok

discord.js@12.5.3

This has been happening with me since months

remote wasp

Can't reproduce the bug on stable. You should open an issue like space said and it would be great if you can also drop a link to your bot's repo if there is one in the issue.

sleek apex

what's the point of this? can't we just have in ./DJSError.js's module.exports Messages = require('./Messages');?

oh

wild flax

Circular references

sleek apex

wait

sleek apex

also is my vscode broken or is the entirety of /src/errors/ used for nothing? oh nevermind i see now

im confused

sleek apex

wish me luck

vagrant holly

👋 Not sure if this got discussed anywhere, but it seems the lockfile in master is now v2, which is Node 15+ by default?

Node 14 still ships with npm 6 by default, and the project claims to be Node 14+, but if you actually want to develop on the project this is not the case and you need Node 15, or to manually update npm in your Node 14 install

sleek apex

most package managers ship the latest version (v16) and brew upgrades all packages whenever you run a command

wild flax

npm 6 can work with v2 of the lockfile, and the library still works with 14

vagrant holly

Off the top of my head, won't npm 6 overwrite the lockfile back to v1 if deps are modified?

wild flax

maybe

vagrant holly

Would it not make sense to at least define npm >= 7 in engines.npm in package.json?

wild flax

maybe

elder reef

Not sure where else to put this... the invite in the org's description is outdated

outer night

it still works

hallow ocean

Yes, it just doesn't use the vanity url so it's fine if it isn't changed

wild flax

Maybe refresh your browser

fringe star
remote wasp

yeah

gusty tide

Hey, apologies if this is the incorrect channel to ask such a question, but can someone here give me an idea as to when v13 is released, as it apparently has support for the new slash commands? Thanks!

real jetty
copper laurel
patent needle
strange igloo
patent needle

Ah alright

raven juniper
strange igloo

I meant that it hasn't been released in stable yet

grave tree

What gateway V is using djs v12

wild flax

7

grave tree

Oh ok

outer raven

I'd like to propose a change to the docs to make it so that the "Default" column in the parameters table displays the default value set by the discord API, even if the code itself doesn't have any default values. This way it would be easier to understand what would happen when leaving certain optional parameters out. Would this be a good idea?

unique axle

no, re-documenting things is almost always a terrible idea, because upstream docs changes could very easily make the information displayed wrong

outer raven

It wouldn't be re-documenting as the current docs don't link to the respective page on the Discord API docs

wild flax

our docs represent our wrapper, not the discord api

outer raven

what I'm saying is that currently it's quite hard to find the default value for stuff because most are left empty

wild flax

so the default values are the ones in our code, not from the discord api

outer raven

ye that's how it works atm, but why not change it?

wild flax

if we start mix-matching it it would become quite confusing

also we would need to keep it up2date with discords docs

thats also why we don't document what perms you need for certain actions

outer raven

wouldn't be too hard, I doubt they'd change values all the time and if they did it would be rarely

wild flax

what about the first point

outer raven

I don't see how it would be confusing

wild flax

how are you going to discern what default value is ours or theirs

outer raven

you're not because it doesn't really matter in the end

if you have a default value on the code, you use that on the docs, otherwise you use the one from the API

wild flax

yeah it does, we have functions and methods that have default parameters that have nothing to do with discord

outer raven

yeah I know that's what I'm saying

we'd prioritise the default values used in the code

wild flax

so how do you document 2 default values then

outer raven

you don't, you just document the one used by the library

because that one will always have priority in the code over the one the API uses

wild flax

and what happens to the one the api uses (just in case they are different)

now you decide arbitrarily what to use when

outer raven

it's ignored because from the moment you have a default value in the library, the default value given by the api is ignored

wild flax

"oh no this isnt a default from discord, this is ours"

"oh no this is a default from us, not discord"

outer raven

it doesn't matter to the user, they just need to know what value is gonna be put there if they don't specify that parameter

and we'd put the one with the highest priority

wild flax

yeah, the user can just open the discord api docs too for that

outer raven

which is library > api

yeah but the discord api docs are way harder to navigate than the djs ones and its easier to have it all in one place

because saying there is no default value is essentially misleading because the result you will see on discord will have a default value

wild flax

its not misleading

we don't claim that the discord api doesnt have defaults

its just that we don't have any in the library

outer raven

yeah but if the library is an API wrapper, it should also document what the API does in certain situations

unique axle

hmm.. no

wild flax

no, not really

thats what the API documentation is for

outer raven

but it's easier to have everything in one place

wild flax

yeah, thats fair

doesn't mean we are doing it

outer raven

why not make the docs more accessible instead of telling users to go search in a million places?

outer raven
wild flax

the docs are plenty accessible

downgrading their value simply because we don't do this one thing doesn't make them any less good

we still have, to this date, the best auto generated library documentation

outer raven

that is true but there's always room for improvement

wild flax

improvement is great, more things to maintain is not

outer raven

If any default value changes, people will most likely hear about it quickly from the discord staff themselves and make a PR to change it

wild flax

just like how interactions were released and we needed 2 PRs for it and a 6 month delay right?

outer raven

I see PRs changing little things that the API changed all the time

wild flax

I can't imagine that for something thats even more hidden and less important

outer raven

and I don't think that delay was because people didn't know about them

wild flax

I say that specifically because interactions could have been done in a 2 week window

No, it was because no one was simply doing it. We don't get a lot of contributions for things the API changes

outer raven

but their release was also quite weird and a lot of things changed

wild flax

That was also not the reason, this discussion is pointless overall

As it stands right now we arent doing it

outer raven

fair enough I guess, I'd be willing to do it if you change your mind in the future

outer raven

you can technically put any type on an embed description but objects and maps will look weird and I doubt people would want to put that intentionally so would it be possible to change the embed description type to string | number | string[] | number[]? It's currently set to any

ornate topaz

the fact that they look weird doesn't mean that they can't be passed

outer raven

Yeah I'm saying they can but no one would want to put [object Object] in an embed description and it would help them identify errors if they do this accidentally (in TS of course)

ornate topaz
slate nacelle
outer raven

that's different because message.member.toString() exists

outer raven
ornate topaz

well, so why not add all other variants like GuildMember | User | Channel to that huge type as well

outer raven

you don't need to because it runs toString() automatically

ornate topaz

but your type would mean that it cannot be passed there at all

i didn't pass a member.toString(), i passed member

outer raven

but since the toString() function exists, I think ts doesn't detect that as an error

ornate topaz

which type are you talking about

the param to setDescription?

outer raven

yep

ornate topaz

why the hell would it care about completely random properties on passed object if the object itself doesn't fit

outer raven

the only reason why it throws an error is because of the number type, but it is assignable to a string type

unique axle
outer raven

wdym?

is that gonna change?

unique axle

space literally linked it

outer raven

I didn't see that it would require you to run the function

unique axle

we aren't doing any string conversion for you anymore, you have to actually provide a proper string (once that lands)

outer raven

why tho :/

ornate topaz

you can't really do const x: string | GuildMember and use x only as a string. if x would be GuildMember, it would require you to call the toString() manually every time, it won't call it for you.

it doesn't matter that it has a function to resolve into string

it literally isn't a string

outer raven

it does though

I just tested it

ornate topaz
outer raven

yeah it doesn't do that

it only conflicts with number, not string

ornate topaz

what doesn't do that

outer raven

if you have a string type and pass a GuildMember, it works

ornate topaz

what works

outer raven

passing a GuildMember object

slate nacelle

Yes, because we forcefully resolve everything you (don't) pass into a string.

outer raven

yeah and TS doesn't detect that as an error

so its all good

but it still won't work due to the number type, which is stupid

and for some reason passing a number to a string errors

ornate topaz

am i missing something

outer raven

hmm I didn't test it like that, I modified the StringResolvable type to string | string[] and it worked

ornate topaz

the c literally isn't a string, otherwise the type would be useless

so anyway, the method works (atm) because it resolves

but you narrowing the type from any to your string | number | string[] | number would mean that you cannot pass anything that you currently can

outer raven

yeah I know I believe that would be benefitial

ornate topaz

if the change that requires strings lands, yes

right now? not really, as the method handles passing some objects (some as in literally anything that has a toString(), so pretty much everything)

technically needing to list them all in the type

unique axle

since the type is literally going into the bin, i'm not sure what to improve about it and why

outer raven

well yeah i wasnt aware

thanks for the explanation

remote wasp
vivid field

Looks like another remnant of the webpack build support

remote wasp

hmm, that's what I was suspecting too, just wanted to be sure. Thanks 👍

haughty anchor

Any word on when library will be brought in line with button update?

unique axle
haughty anchor

Thanks sorry

wild flax

It’s maybe been 24h since release. At least a little bit of patience would be appreciated

haughty anchor

Im not trying to hurry anyone I was just wondering what the general timeframes are normally like, this is the first API change that will actually affect what I work on for the time being

The activity on that PR is nuts

Is better news than I was hoping for

copper laurel

I've just marked it as ready for review and testing now

haughty anchor

Yeah i saw you are a madman

copper laurel

There was a lot of base structure and classes to be built - but it should make future components a lot easier

haughty anchor

Thankyou for your service

chef

loud jayBOT
copper laurel

Buttons are now ready for review and testing, but may not be stable

fair yarrow

nice

copper laurel

Just pushed breaking changes to it, FYI anyone already using it

sage owl

To someone who actually works in the industry, “When it’s ready” is such a cop out response. It allows Scope Creep to become a real problem which v13 is starting to show signs of. So I think many of us, who are relying on new features being implemented, would like a little more effort in that part of the department. We understand things come up and things get pushed back but no reason to set goals for your releases. Setting Target dates is important in any software lifecycle, especially when people actively use your codebase and count on these new features. Anyway, this is just some words of advice from someone who works in the industry as a Software Engineer. Take it or leave, ignore if you like.

real jetty

A key thing to keep in mind is that you can still use it if you really want it. Installing from master does work and you can choose from which commit.

manic zinc

@sage owl have you considered using your professional experience to help move the project forward so it gets done faster? I'm sure the team would appreciate the addition of a Software Engineer helping out on the project!

outer raven
outer raven

Also for the server avatar feature, should the guildmember object have a displayAvatarURL() function or just the avatarURL() and avatar properties?

copper laurel

What would the defaulting be

Server avatar > Global avatar > default Discord avatar?

fair yarrow

monbrey

attaching Imge file in embed in interaction response won't work

ornate topaz

Fairly sure it's on API on this one, someone brought that up before

@outer raven typescript also doesn't follow semver, but that's different topic

fair yarrow

Ohk

but can i attach files on interaction reponse

copper laurel

Only in the edits

Dunno why

copper laurel
sage owl To someone who actually works in the industry, “When it’s ready” is such a cop o...

We know we have a huge community that are excited for new features to be released, but Evie is right. A lot of the staff, myself included, and particularly the maintainers are professionals, we do work in the industry - that's exactly why we don't have the time to manage this like some sort of enterprise scale product.

We don't have a dedicated testing team, we don't do sprints, and the majority of the people who use the library are, to put it bluntly, amateurs. They don't understand how to handle semver major breaking releases.

That ties into a lot of decision making around what goes into a major release, how we can be prepared to support it (eg v13 changes guide) and ultimately, when we are ready to release it.

I get that "when it's ready" might seem flippant and dismissive, but setting a date that we don't meet is pointless. Alternatively pushing out a buggy mess that hasn't been properly tested just because a date was set is even worse.

fair yarrow

oh k so no way to attach a file nor show in embed

solemn oyster

Also keep in mind Discord also likes to introduce breaking changes from time to time, which screws our release cycle at times

@fair yarrow you might want to use the support channels for this, this one is reserved for discussions regarding the library

echo fog

when will message components be implemented?

solemn oyster

There's a PR for it

echo fog

ah

nice

clever spoke

?buttons

rich iglooBOT
clever spoke

worked on there

echo fog

epic

fair yarrow
outer raven

I made the PR meanwhile and added the display avatar url method which sends the server avatar or the user’s display avatar

outer raven
ornate topaz

feek

strange igloo

Major . Minor . Patch*

Major for breaking changes
Minor for features
Patch for bug fixes basically

strange igloo
wild flax

Very unlikely

outer raven

I don't see why they wouldn't be able to

wild flax

What lol

strange igloo

Since it's going to be restricted to their $10 Nitro apparently, I don't think they'll allow that

wild flax

Can’t wait for some mods to set my image to something I don’t want

ruby terrace

I'd find it unlikely they unlock that feature to bots

wild flax

You really can’t see how this isn’t abuseable?

ruby terrace

while those "self host" bots would love it, that type of data storage for large bots seems like a bad idea

wild flax

Oh, the idea of the bot having one

Puh.

outer raven
wild flax

Not even sure that’s smart

outer raven
wild flax

I can totally see someone setting a different avatar on their 300k guilds

pepeLaugh

outer raven

well yeah

ruby terrace

yeah....that

wild flax
outer raven

yeah true

well ignore that comment then lol

strange igloo

About Nitro perks for bots, I don't think they'll grant perks from the $10 one
Not even sure if they might allow any more perk from any of both Nitros (well yes animated avatar is an example)

copper laurel

bots have global emojis

No other nitro features

tacit crypt

IIRC they get upload limit based on the guild's boost level

ornate topaz

like every other member of that guild, so irrelevant

oak quail

any chance of changing CommandInteraction#options to a collection with the option name?

copper laurel

I dont think discord enforces uniqueness

real jetty

The options are send in order as you'd expect them, so i appreciate the array here

copper laurel

oh it does enforce

remote wasp

It's much better to work with collection, so if it's possible (which it should as names are unique) I'm all in for this 👍

strong heath

will we have buttons with author only funtion?

tacit crypt

Buttons are added just like discord did. You can handle author-only buttons yourself

outer raven

Is it possible to link a method in the docs by using (for example) {@link User#displayAvatarURL}?

wild flax

youd use @see for that

just search the repo for it

outer raven

talking about something inside a sentence

@see is a parameter in jsdoc

wild flax

in a sentence probably not

outer raven

oh I found it somewhere else, seems like it is possible

obtuse hazel

Why does d.js don't support editing an attachment?

wild flax

If your question is why we don’t support a certain feature, the answer is most likely because no one PRd it yet. This doesn’t only apply to this, but pretty much every “why doesn’t d.js support this”

copper laurel

But we do support it don't we? That PR was changed to allow you to edit with a new attachments array

ruby terrace

haven't tested it, looks like it should be possible, but the jsdoc string needs edits if so

remote wasp

@real jetty You still haven't explained why are you confused about #5705. If you have any questions about it then ask. Your reaction just made others to react too without leaving any meaningful feedback. So, It would be really helpful if we can discuss what needs to be changed in the PR using words

obtuse hazel

Oh well it'll be hard to answer if you don't ping me

remote wasp

@obtuse hazel How am I supposed to know you're not using your main. Anyway, that's not the point here. So, what is it that you found to be confusing?

obtuse hazel

I removed it, as I thought it would be annoying if the order matters but you can still convert it to an array and the order will be the same. Sorry for this

tender surge

const user = await message.client.users.fetch(userId)

user_id: Value "a b c" is not snowflake.
would it be an idea to provide a isSnowflake utility function when dealing with user input?

loud jayBOT
tender surge

Hm alright thanks

outer raven

What do you guys think of the idea of adding a type called TextBaseChannel and VoiceBasedChannel so that we don't have to update our code so often when a new channel type is added

TextBasedChannel would be TextChannel, DMChannel, NewsChannel and (soon) ThreadChannel, while VoiceBasedChannel would be VoiceChannel and StageChannel

would only affect typescript users of course since only the typings would be changed, no actual code changes

ornate topaz

isn't that what isText() does

outer raven

that's not what I wanna do though, I wanna set the type of a channel when fetching it or something similar

instead of doing as DMChannel | TextChannel | NewsChannel I'd do as TextBasedChannel

ornate topaz

when you fetch, you don't do the first one at all, library does it for you

and what would that TextBasedChannel even be? a type union? that still doesn't solve the problem of having to check if it's a dm or not, or if it's a specific type

not sure what would even be the use case for such thing. isText() can be used to somewhat narrow the type to channels you can for example send to. sure, there isn't stuff like isGuild or isDM but you can check that yourself with one line (and if done properly, also narrows the type further)

outer raven
ornate topaz

...and what exactly would be the benefit?

  property 'guild' does not exist on type DMChannel | TextChannel | NewsChannel | ThreadChannel | WhateverElseChannel
    property 'guild' does not exist on type DMChannel```

if you are fetching/getting by id, you can literally just cast to a correct type already.
if you are requiring a user input for an id, you can either assume that if you ask for "log channel" you didn't get a voice channel, or just check while you're in that form, and then just have only TextChannels anyway

there is 0 point in casting it to multiple types and then complaining that it has multiple types

and your union would be exactly that - multiple types that are just a fairly big subset of a Channel, and still need narrowing

true, there is a hassle every time new channel type is added, but there's nothing stopping us from adding the type, especially considering both how infrequently new types are added, and that it's almost sure that new channel type will have new features that would have to be added anyway

outer raven

It's as simple as adding any new channel types to the proper type declaration and updating them as new ones are added.

ornate topaz

and what would be the use case for such type

copper laurel

I still don't see how this is different to isText(), which asserts that the generic Channel is one of the three. Type guarding functions are better than casting wherever possible. If the argument is for a way to easily cast what a guard already does for you, I don't agree

keen cosmos

Sorry for my review, I thought we used flatMap elsewhere as well, should've checked first

copper laurel

No worries, it was worth looking at because I forgot it existed

wild flax

@remote wasp im not a huge fan of this pr

because we slowly want to migrate over to discord-api-types

how were the types imported? and what was the error?

remote wasp

yeah, I also don't like this fix myself but the issue can't be fixed without it.

remote wasp

It's like djs doesn't even export them

I have reloaded VSCode and have the latest commit installed (that re-exports the snowflake)

slate nacelle

Maybe we should export all of those interfaces and everything that's actually exported then.

remote wasp

yeah that could be a fix that I was thinking too but I had a question in my mind too, which is:

There might be a correct way to re-export types which might fix it but one thing that I don't get is, why are we re-exporting Snowflake. We use a lot of other typings package like for node, ws and other but don't re-export anything. The end-user installs those for themselves and use it. Shouldn't they install discord-api-types and import Snowflake from it?

ornate topaz

couldn't that be explained with the fact that snowflakes are basically the most common thing you use to interact with library?

wouldn't make sense to make a type for getting stuff from every single collection/manager and not reexport it in the lib

for the node typings, im fairly sure almost every project installs those anyway, and i think ws ones changed from being required to install manually to being installed by lib

wild flax

it depends

usually yes

I would agree if it wasn't such a common type

also uh, it doesn't look like the Snowflake export is at fault here then

if we don't export those interfaces, you usually can't import them. I have no idea why it worked before lol

vivid field

Because we use declare module

wild flax

whens your PR making a return

vivid field

I can do it now ¯_(ツ)_/¯

That'll break basically every other pr though

Or at least the ones that touch typings

wild flax

yeah we prolly want buttons first then

and then your pr

worthy coral
unique axle

yeah! discord can also randomly send typing events to voice channels

worthy coral

amandathink

That is very strange. I've never heard of that or experienced that before. (probably just lucky?)

wild flax

If you use a library it won't really happen to you

But if you use the raw api it can for sure happen

and you need to handle it

worthy coral

Working on my own and taking inspiration from Discord.js which is why I asked. I will definitely keep these in mind, though. Very helpful to know

unique axle

this is the type of thing you find out when it happens, you investigate the issue, nothing makes sense, you ask very carefully, because it's a horrendous question and then get
"oh yeah, you need to handle this" as response kek

worthy coral

Not fun so far. Send help hidethepain

copper laurel
ruby terrace

I'm adding guildID to many of the methods in ApplicationCommandManager rn so you can manage commands on guilds that don't have a bot user / aren't cached and I was noticing something.
Would it be desirable to turn all of the manager#fetch method parameters into an object (for those applicable, still accepting resolvables as the only parameter) for consistency?

Current managers that use this structuring are ApplicationCommandManager (with my PR, and therefore GuildApplicationCommandManager), GuidBanManager, GuildManager, GuildMemberManager, ReactionUserManager
Those that do not and would be changed: ChannelManager, GuildEmojiManager, MessageManager, RoleManager, UserManager

copper laurel

You mean like MessageManager#fetch({ message: MessageResolvable }) or something?

Because I really hate that

ruby terrace

No, it would still take bare MessageResolvable, but if you want to not cache or force, you need { message: MessageResolvable, force: true }

granted, it could be fetch(message: MessageResolvable, options: BaseFetchOptions )

or something

copper laurel

okay that I dont hate

the second one

fetch(message: MessageResolvable, options: BaseFetchOptions)

ruby terrace

yeah, okay, need to change some of those other ones to follow that style then too, but I think that would be better for consistency

hushed sedge

Idk if suggestions are allowed here, its the most relevant channel i could find

wild flax

We aren't discord.

We have no control over what is allowed and what not, or what will be.

hushed sedge

Oh, alright. I didnt know that.

well i knew that

but i thought you had the ability to change what you could send

to an extent of course

copper laurel

You probably can send it, and then the API will reject it or the Discord client wont render it

hushed sedge

I see. Thanks for clearing that up.

copper laurel 
public on<K extends keyof ClientEvents>(event: K, listener: (...args: ClientEvents[K]) => Awaited<void>): this;
``` This change to use `Awaited<void>` in the Client event listeners effectively prevents you from returning anything from the function, eg
```diff
- return message.channel.send("..."); // will error
+ message.channel.send("...");
+ return;``` Is this intentional, and required for some reason? Could we not do something more generic like `Awaited<unknown>` because it doesn't matter what a listener returns (or does it)?
ruby terrace

Hmm, EventEmitter doesn't care about return values, in fact there is no way to get them. The @types for EventEmitter define it as void, but I do think unknown would be better.

solemn oyster

If return message.channel.send("..."); were to throw, you'd get a stack trace, but not the async part, because the caller (EventEmitter) isn't asynchronous

copper laurel

Okay, cool, thanks

solemn oyster

np

strange igloo

Shouldn't the library prevent methods such as VoiceChannel#join to be executed if the client has not subscribed to the required intent? Or complete documentation by saying "to use this method, you will need to subscribe to that intent"?
Several people have been running into issues related to these (the example I have in mind is VoiceChannel#join throwing connection timeout).

wild flax

Probably not, no.

We don't do it for anything else.

slate nacelle
wild flax

Perms yeah, but we never check for explicit intents no?

slate nacelle

No, those didn't exist back then after all.

Point is to avoid timing out, since that's a poor (and slow) error.

remote wasp

yeah, the error takes 15 seconds to throw. We can make it quick by checking for the required intent (GUILD_VOICE_STATES?) in the VoiceChannel#join and throw if not found. We do check for intents while logging in so I don't know why we can't do this too. Now this takes me to the question:
Why do we set ClientOptions#intents to the resolved bigint instead of an intents object?

slate nacelle

I don't think there is a reason for that.

remote wasp

hmm, then we should make it an intents object, then we can easily check whether the intents user chose have an intent. Currently we will have to make a new intents object from the bigint everytime we wanna check

wild flax

I'm not sure if checking for specific intents is the right path here.

Theres a fine line between good user experience and "well thats kind of your own fault, the vacuum cleaner won't put dirt into the bag if you dont have a bag"

It will still turn on however, and leave a big mess

PepeLaugh

remote wasp

yeah, I understand. It would have been the user's problem to fix if the error would be quick & clear but it takes 15 seconds for the error to throw and on the UI side the bot still shows in the VC. The error also doesn't help: "cannot make connection within 15 sec...something..." Which mean that the user has no idea what's causing it

wild flax

Also its a bit problematic, because technically speaking we want to move away from our internal voice code anyway and mostly delegate things to @discordjs/voice