#docs-website
1 messages · Page 14 of 1
Just hide a 5x5 image of Paper's icon in the top-right
get rekt strok
ezgif is just goated
But where CLI tool?????
yeah but less frames
if you use 33fps (which you can) it looks better
but it is still limited to 800px width
instead of 1920px
That's horrible
MY WIFI DIED
It's way too low fps
For a 180hz gamer as me that is unacceptable!!!
good question
all they allow
I always use webp in websites
I am working on a platform that allows users to upload files, I am also converting them all to webp on upload
storing them only as webp
sharp converts most of our images to webp but it doesn't for gifs I think
wow, how is that so small
did ya guys try AV1?
I also use FFmpeg but it creates huge file sizes
Somebody pls recommend a cli tool to create webp stuff
Too much effort to select the codec in ffmpeg
I also used vert.sh to make it into a gif and it's roughly the same size so idk what yall's ffmpeg settings are
Default settings 
No, that's not a cli
on top
maybe it has a rest endpoint
alright will convert
then it is a cli tool 
this is barely visible
perfect
just tried it, not worth it, became 2.1mb
alright, will make webp images then 👍
Apart from nuking the quality and fps, what does ezgif to differently than ffmpeg does
also, tried ImageMagick?
ezgif likely uses imagemagick yea
Read about it but never used
yeah imma take a look into imagemagick after I eat breakfast
it's 4pm and I still haven't eaten anything
was too busy spreading misinformation on the internet
not if this is true
if you ask me they just optimized their commands
Is this too small?
Okay the first round of pictures is up and I'm pretty sure ezgif blocked me for some time 
I’ve been silently following the process and I must say it has been fun and thanks for your contribution
@twilit fog
for some reason i have been following this too 😄
hey, it's been really fun for me too haha, I decompiled client code and found some really neat stuff, INCLUDING a paper bug :D
@twilit fog I mean the videos just don't loop 
It runs once and then never again
At least for me
yeah same for me
but as I said in the PR, works on my machine so this is really strange
The thing is you have to explicitely encode the webp so that it loops. You might just be using a video viewer with ignores the native video settings for looping
Or your browers might auto-loop them for you if you only view the raw file, but not embeded
I view them in the browser because Windows is stinky and doesn't support them
but it's the same browser?
what is going on?
^^
i see
I guess I have to figure out how to encode this with ffmpeg, sinc ezgif doesn't do it
man how does ezgif do this ...., I can't get nowhere close to that
omg I figured it out
I was settting the encoder to libwebp and found a stackoverflow answer that said "remove that lol, let ffmpeg decide what to use"
106kB now
here is what I use: ffmpeg -i input.mp4 -ss 4 -t 2.5 -filter:v "fps=30" -loop 0 -an -vsync 0 -s 1280:720 output.webp
works well
neat, I can make them 720p and still have them in kB
just tell me, is this fine? (the direction) I've changed the color to yelow but am not sure if the direction is okay
nvm changed it
Tbh I almost don't see it

^ those test should be done in a blank world or in the sky especially for small particles
guess you can put a block in the middle of nothing
thats too obvious 
oh thats a great idea
but yea, visibility obvious over funny paper reference xD
eh I still would agree that the background makes this quite difficult 
fair enough
there was an issue for gifs not looping in astro but webp shouldn't have problems?
need dark/night mode xD
thankfully I didn't use fireflies anywhere
this reminds me that I gotta restart work on Hangar after tomorrow's exam...
well... after this weekend, cause ya boi gonna be playing BF6
is there any way to test locally with embeded?
pnpm dev and look at it?
Works on my machine ™
doesn't work on preview
I have a feeling astro is just straight up nuking it, I'm sure if I put it in public it will work
what's not working
webp is not looping
I'm 99% sure it's Astro's or something else's fault because my local run doesn't use the transformed one but rather url

guess I'll use public
use the img tag + mdx import for now, the image service is stupid
Love to hear it
hmm interesting, yeah sucks that the width is limited to 800px
but welp
overall looks nice
good job
Wait what, what browser is that, looks fine on firefox
Safari
Ah
"fun" 
Thank god my job does not involve web dev
How am I supposed to test on Safari ahhhh
I mean ur browser can emulate the mobile resolution and check that
Can it emulate safari?
If I try to remove edge from my taskbar it will magically reappear like a ghost
It’s a sign that it must be the best browser
Mine was really blurry too. Ping me if u want me to check it
Could be two things really, either its the canvas operations or the image loading
uh? I uninstalled edge without any issues
thought they had made it super easy to do so
also, agreed, no one cares about macos users
It’s a managed device so I figure they force it to be there
safari is so ass at image scaling ffs
I really hope copilot is wrong:
When you draw a canvas onto another canvas (as in your code), Safari often ignores the smoothing setting and applies smoothing anyway.
Imo, its ok if it only breaks on safari lol
Smh EterNity hates accessibility
No one talked about edge here
It’s the better browser
Nah
True
If you use any of them u will get used to it
I use Vivaldi, it works very nicely for me
looks fine on safari for macbook wrong browser
Yeah, it's chromium based
It's simple and it works well for me
Looks clean enough
It's fairly customizable
Are extensions compatible with like chrome or Firefox?
i mean its chromium based
Doesn’t necessarily mean it works out the box
nvm i was not using safari to test it 
The one annoying part is that downloads and some other stuff is in this sidebar here, but you get used to it
It even got pretty good email integration, so that's a plus for sure
deeply disappointed the note in the GIF above does not change color when you move the slider
💀
Feel free to PR it 
what
Eh. Sometimes I like an app to stay in its lane
Nah I like it. Saves me having to open my mails app every once in a while
That’s fair, I just have outlook open in the background normally
Fair enough, I don't do that usually at all
ive got thunderbird, as thats the only email app i know besides outlook
I used to use thunderbird, but vivaldi mail is more convenient for me
okay so I think I found the solution hopefully? but I will commit it tommorow
canvasContext.webkitImageSmoothingEnabled = false should potentially fix the issue
I don't understand why it's still vendor specific
Apple doing apple things
there is also mozImageSmoothingEnabled, I have no idea what moz is tho. Mozilla?
there's more....
Probably Mozilla
it was a sharp regression, it's fixed now
it's baseline, so no need to use the vendor property
Then why doesnt it work on safari?
Is copilot right that safari just doesnt care?
does the vendor-prefixed version work?
haven't tried yet
one sec
I mean the IDE is crying that it doesn't exist but there aren't any errors
Alright I changed the canvas logic a bit, @neon epoch it should work on safari now
is that safari?
Yes
okay thanks, it's better than it was
will look into it
actually it's not, it's the correct size and everything but your browser is doing AA for some reason
the top and bottom seem cropped because AA being applied there isn't visible
@vocal halo if you move the slider, does it get better?
ok thanks
Hopefully this should do it now, those of you with Safari I ask for testing :D
🎉
Copilot was really helpful, he didn't write the nicest code (nor working), I just read what he wrote and did it better. First I tried googling but I didn't find anything really
Fire
The proper use case of AI with human supervision. Wow.
https://privatech38-particles.papermc-docs.pages.dev/paper/dev/particles/
I really hope this is it...
Better call some fire fighters 🚒 🧑🚒
I create more fires (bugs) than I put out (fix)
why even include negative offset in that slider? seems to be the same colors as the positive one?
well, same as in offset by 1 i.e. -0.75 and .25 are both red?
Because that’s probably the possible range?
correct, it's mostly just so people have visual confirmation that there is pattern repetition
@pliant canopy can you address the MM comment on the dialog PR and switch to webp for the videos
I'd like to merge it sooner than later
Sure
@twilit fog out of curiosity, what did you end up doing for the webp videos? Did you just run it through ezgif or did you manage to figure out some CLI way?
scorp linked it to me, but for me it produces rather.... unsatisfactory results
Unsatisfactory?
relatively similar
But it just looks bad
Imma try updating my ffmpeg, maybe it is just wildly outdated, who knows 
Possibly
maybe force the framerate?
Tried that too
idk lemme try that in a sec
Do you maybe wanna try to convert it?
Maybe it is really just my PC screwing with it
seems to be also happening on my end
@pliant canopy could it be the bitrate? Yours is really low, I have 2000 kB/s while yours only has 200
Who knows 
try rerecording with higher bitrate in OBS
I managed to make it look not entirely shit but still relatively fine
Problem solved
what did you do?
I used gifski to make a gif first and afterwards converted that to webp
hmm
if you ask me its got to do with bitrate
that's true but who knows
It's gif now. I kept the big videos as mp4 as they don't need to be autoplayed and are hidden by default anyways
Doesnt astro just convert gifs to webp anyway?
Idk and I don't care, as long as I don't even have to touch webp again I am happy
I think your hatred for webp is unrelated to webp haha
in that case it's my hatred for libwebp or whatever the lib name is
Okay yeah I agree it took me some time to get the wanted result
Just dumping this here for me very quickly: ```
ffmpeg -i input.mkv -f yuv4mpegpipe - | gifski -o output.gif -
ffmpeg -i output.gif output.webp
Just so I don't forget it
Could theoretically pin it, but eh idk if that is worth a pin
https://docs.papermc.io/paper/dev/menu-type-api/#what-are-inventory-views, the last link is broken Strokkur
huh
Uhhh, woops
Weird...
I would have expected that scorp's thing catches that
I placed a dot where I should've placed a colon
fixed
oh wth main is not building. Bruh
fixed in the wrong branch
Yeah I know, I was about to merge this one in
Why is main failing to deploy
What is this error, I have never seen that before
Did github actions run out of internet or what
the runner is just fucked up, I restarted the job
Aight
btw could you make it so that if you make a jd link which looks like this, it get's flagged? jd:paper.org....
That's what had happened there
aka invalid project
it's fixed on my macbook since early july
Lmao
hmm another weird link
Darnit, that's why I hate merging
(Not really, I am just bad at it)
Thank you for pointing it out though 
Before I push my patch, do you happen to anything else out of the ordinary?
Just so it happens all in one
no not on that page i haven't reviewed https://github.com/PaperMC/docs/commit/1c2cf26748c7f1d1eebdff4028da38dbb06eb6a1 yet but you can push really
Btw a heads up: I am planning to also merge #631, so if you could also look over that one whilst you are at it, I would appreciate it

Probably going to look twice at the jd links for this one anyways, but 4 eyes are better than 2
"perquisites" 😭
Did I misspell that
prerequisites
I don't see any documentation on what MultilineOptions.create(int, int) does. Maybe I didn't see her.
an inventory category could make sense
and maybe the dialog too? it is a kind of menu of some sort but not sure about that one
I think that would make sense, but I wanna wait on scorp's opinion for that one
dialogs are not inventories but the rest yes
I meant in a menu way
I guess they are user interfaces
yes
could also guide devs to how to make user interfaces
Tbh I am not sure if that is in-scope for the docs
hey uh, the button for lifecycle api folder is not visible in the https://papermc-feat-datapack-discov.papermc-docs.pages.dev/paper/dev/
yeah the slug in lifecycle.md should be updated
make sure to also update the link in datapacks too
the slug shouldn't be updated, since that warrants a redirect
slugs in general don't correspond to the file structure, so it's fine to leave it for the existing page
I'd set the sidebar label to "Introduction" or similar for the lifecycle api page though
sidebar:
label: Introduction
in frontmatter
Well, I will have to change the lifecycle page's slug anyways, but it shouldn't kill of any links as instead it will just contain the index
Actually nvm it would kill links and I don't need to change it after all
scorp, is there any way to disable the image optimizer for PR preview builds? It takes horrendously long and I hate waiting
It adds a whole 2.5 minutes to the build time, which is just unacceptable
A build time of 186 seconds, where 154 seonds are spend optimizing images lmao
no, preview builds are supposed to be production-grade builds
set up a cache on the action
Hm, fair, but it's still kinda annoying
Works wonderfully (a build takes 32s now), thought I am slightly concerned that I might be caching a bit too aggressively. Currently doing the dist, node_modules, .cache, and .astro folders. Should I remove some of them? Most notably, I assume the node_modules one is not actually reqiuired because we have the node cache setup anyways?
Also sorry, that was not meant to ping, mb
Ended up removing the node caching and just doing the caching manually, because when I removed node_modules it didn't cache it fully 
Works wonderfully, at least I think so
Just a heads up, scorp, for some reason it is not caching anything on the main branch. It seems to work fine on PR branches, but not on main. It does say Cache restored successfully in the actions logs, but then proceeds to not use the cached images
. I am out of ideas, so I am letting you deal with it in case you know what is going on.
you should only really cache dist/_astro/ since caching the rest will possibly include stale pages if we ever remove something
When I do that it seems that sharp does not correctly cache the images
Which is really weird, considering they should be in /astro
(testing this out locally rn)
It seems that it stores something inside node_modules. If I remove the node_modules folder, it completely regenerates them, even if they are inside dist/_astro
Yeah it seems to take it from node_modules/.astro/assets lol
then cache that
Yeah will do 🫡
Alright works incredibly well, both in PRs and on main
Finally I can sleep in peace
(I just woke up)
hey would an addition of Flyway and or JDBI chapter be welcome in the Using databases page?
Flyway seems to be problematic because it doesn't use the plugin's classloader and doesn't find the right files
it should be generalized to a note about using the serviceloader with the current thread context class loader
javalin and other libraries suffer from this as well
after wrestling with Flyway I found out the only valid classloader is the one from JavaPlugin.getClass().getClassLoader() since each plugin has its own class loader.
Shouldn't getClassLoader in your plugin class work?
It's protected so can't use outside of that class, but that should be the class loader to use
A full on tutorial for jdbi and flyway seems Overkill, but mentioning them, linking to relevant resources and documenting gotchas like this seems a good idea
YourPlugin.class.getClassLoader() :p
this.getClass().getClassLoader() 
That's not the same method
it should end up in the same result though
Go for it 👍
thx❤️
why did the colors change
Which colors?
download explorer
that was not the color update when "channels" was introduced?
The colors matched the ones Riley picked for Fill
We just haven't seen the beta color so far
not really liking the stable color
^^
Make it unstable then
thoughts on website PR 151?
I would have thought "id" is more correct than name, which I'm inclined to treat as an arbitrary display name string
Tho I think field intention is missing from openapi docs a little, esp as fill isn't open source yet (right?)
Tho maybe introducing SCM URL into fill project metadata wouldn't be the worst?
It needs to be defined somewhere, after all, and I think centrally for all projects makes more sense than possibly n consumers of the api
It did def feel weird
But didn't care enough to comment
I agree that an explicit field is always better than something implicit
If only fill was open source so that I could yell at you to open a PR....
Need like a “day X of minidigger asking for fill to be open source” counter
I mean, I could press the button
I dare you to press the button
aren't you added to the org
Doesn't mean I have access to the repo or if I were to have access to the repo, that I would have access to make the repo public
i moreso meant don't you have access to it then, not actually being able to make it pulbic
Well, no I don't have access to it
Is that visibility, or org ban?
Both 
@sharp pivot btw I got docs#628 open for the PR you just merged. Could you give it a quick glance over if what I wrote there as a description is fine? I feel like it is a bit tangled in-on-itself due to the name change
something along the lines of "Disables an optimization that [...], allowing remote ender dragon respawning to work again." would read better
Yeah, thanks for the suggestion
I think this is good
Thanks again, once the CI pases I'll be merging it
yeah I know
@drowsy zinc I am learning a bit far out of the window with this one, but do we maybe want to work to merge the Adventure docs PR soon™? We could just emitExternalStylesheet: true for the code block bug until it gets fixed (it's never getting fixed lol) and the remaining problems should be not super hard to finalize
what code block bug
It affects a few pages which display dependencies: https://strokkur424-feat-adventure.papermc-docs.pages.dev/adventure/serializer/plain/
See how the code block is malformed?

enabling that would mean having disgusting layout shifts on every page with code blocks
idk I'm not in a hurry to merge it, the current docs work fine
It's still just annoying to have a PR laying around doing kinda nothing
But hmm that certainly is weird
annoying? lol
it's how stuff works
can i see a before and after of the malformed codeblocks? how many pages does it affect?
i'm with strokkur on this one - having a huge content PR being blocked by a visual bug is kinda meh
Before and after is this: https://github.com/withastro/starlight/issues/3207.
It affects about 5 pages I think? Basically almost all of the serializer ones
There are workarounds: a.e. we could just not use a custom component and 1:1 paste in the dependency blocks as markdown blocks, but then those pages get +50 lines each lol
oh no 50 lines
what mbax said
do it as a temporary solution and then create an issue to track it
leave a comment on each of those pages that leads to the newly created issue
Oh hey it's getting fixed https://github.com/expressive-code/expressive-code/issues/342#issuecomment-3221323255
No need to add my 50 lines (I haven't gotten around to it yet, that's all)
Yeah so @mighty dirge, I have taken the liberty to rebase the PR so that we can see a few of the recent changes there as well. Basically, I am not a fan of of the details extra CSS. I've been conflicted for a while, but personally I don't think it fits or is necessary. I believe we can make all of the details a bit more visible in general, but I'd consider it out of scope for the PDC PR. If you were to remove that I think we could merge it.
hm...I saw, came and went
well, I will look at it again (tomorrow)
With that being said, @drowsy zinc, opinions on improving the clarity of detail sections as a whole? A.e. a slightly lighter/darker background color in general just so that they aren't fully skipped over if read through quickly?
it's supposed to be just a way to collapse content
if you put important content inside a details tag, that is really on you
such content should be inside an aside or just not collapsed
Why wasn’t it put inside an admonition in the first place?
Quick checking before a plane takes off, does our docs still say permanent block breaking unsupported config includes mushrooms? It was apparently Mojang patched recently.
It doesn’t directly
Whether unbreakable blocks can be broken with Vanilla exploits. This includes bedrock, end portal frames, end portal blocks, and more
Thank you. Turns out the plane has working wifi this time so I can reply sooner. Seems the user in an issue was confused. Maybe a translation issue.
where does this com.velocitypowered.proxy.connection.client.ConnectedPlayer; come from? because I found it in the Ambassador plugin but as for the docs (and my IDE) it doesn't exist
That's an internal Velocity class: https://github.com/PaperMC/Velocity/blob/dev/3.0.0/proxy/src/main/java/com/velocitypowered/proxy/connection/client/ConnectedPlayer.java
And how do I get access to them via the API?
I don't know. I think you'd be better of asking in #velocity-dev. This channel is for the documentation
Yeah ok thanks anyways
can we get a note on the bug fixes page mentioning tripwire duping, and how some setups are patched by other things paper patches but other ones do work with the option
hmm but this is a bug right? like the current implementation of the BlockPlaceEvent is really awful
is the written for version modal supposed to display the version it was made in?
Yeah, it was originally written with 1.21.7 in mind
Meaning the last revision was in 1.21.7
because I personally find it confusing for it to be that way, if that's its intent then something like "Since version" instead
I mean, the reason it is not "since version" is because for example PDC is a concept that's existed for ages. So if there is "Since version 1.21.5", because something is documented there which didn't exist prior to 1.21.5, people might thing, PDC wasn't introduced up until 1.21.5, which is just wrong
But written for version 1.21.5 clearly means that the page was written with 1.21.5 API in mind
It does not, at least I didn’t see it that way neither the guy I was helping yesterday lol
Well, what does it look like to you?
Well, after re-reading yesterday’s conversation, it seems like it was more of an issue with the placeholder being wrong for the registry modification documentation, it uses methods that weren’t there on the version it mentions (1.21.1)
Not sure if that number is supposed to be updated whenever newer methods are used or just leave it as is since the base API is there and just add a note about the method
In that case, somebody might have just put the incorrect version number, but the version written there should be the version where all of the methods exist
Feel free to submit a PR which increments that version number to match the methods used @hearty briar
dont know if this qualifies for here but,
if you go to https://papermc.io/downloads/velocity, and then click on build explorer, you get a client side react exception
Brave 1.81.137 (Official Build) (64-bit)
Chromium: 139.0.7258.158
Linux x64 - Fedora 42 KDE
works for me on Firefox 142.0.1, Windows 11
Amy extensions? Works fine for me in Brave 1.81.137, Win11
nothing relevant, only keepassxc-browser and plasma integration
aha
it works only the first time
if you click on build explorer, then go back in history to the download page, then click build explorer again
thats when it happens
issue or pull request to https://github.com/PaperMC/website, please
I have two question.
in this section (https://docs.papermc.io/paper/dev/getting-started/paper-plugins/) how we register some details for a shop plugin unless the shop plugin is not loaded?
# Now, we are going to register some details for a shop plugin
# So the shop plugin should load after our plugin
SuperShopsXUnlimited:
load: AFTER
required: false
and next question, if I can omit this dependency node because I need just each key's defaults value, is correct that each node appears as follows?
YetAnotherPlugin:
SuperShopsXUnlimited:
SuperDuperTacoParty:
I found its answer
upper case is caused by omission of server section. it is real case I want to see.
You give it no keys in the map, so {}.
This isn't about the docs website itself though, #paper-dev
If someone feels bored and wants to help with this: https://github.com/PaperMC/website/issues/154
I think I just found a new side quest 
(Unless somebody gets to it first, I am not taking priority here)
use https://forums.papermc.io/threads/1-21-7.1635/ (and all previous annoucements) as an example of the type of content that would be there
What should be the base? main or Astro?
oh ok
why did you guys give up on astro, out of curiosity
seemed like quite a bit was already refactored for it
We have interactivety on our site
and astros solution for that was that either we wrote JS for it or we imported a component library
not fond of maintaining vanilla JS, and pulling in a component library to deal with the huge escape hatches we need felt kinda cludgy
interesting, makes me wonder of the appeal for astro if not for a site like paper's where even if there's interactivity, it isn't the main content
I guess just not having a good ecosystem for the things you guys are doing plays a role on it, sad to see that given it looks like a promising technology
There was a level of upset over the idea of us migrating to astro AND pulling in a component library and it just felt like the thing had gotten stuck decision wise of how we would deal with that
Stuff like the download page is fun because stuff like the download box dropdown stuff becomes a headache when you don't have somewhere nice you can just store the selected state and disperse of it in one place
would of likely needed to deal with every single dropdown having its own optional "take up the entire screen" box to deal with the interactivety where
@opal flare btw I wanted to ask how sophisticated you wanted the announcement/blog feature to be. Is something stupid, but straight-forward like "just read all files in directory X in build-time and construct some nice components statically" about what you expected or do you want a whole thing of "dynamically create a post on the website, and post it, and posts are fetched dynamically on clients, and it is all fancy and auth and woo"
There is a very crude line between the easy approach and the complex approach, at least if it wouldn't use some external library. Haven't looked at that yet, but just wanted to get your opinion already anyways
-# I suck at explaining, it is 11pm, but I think you know what I mean, sorry for that 
I think easiest and fastest might be to fetch from an endpoint? That way you have fast delivery and also not too much effort in terms of making an entire solution to just publish announcements
I mean, you still need a way to post to that endpoint
That’s not the hard part. That can be a custom script with an auth token that probably already exist somewhere on the machines of people
Like, I don’t think they asked for the community to build such a system if they want a complete blogging solution
There is a lot which is possible. I kinda wanna hear what riley had in mind before I start implementing a super cool system
Yeah I mean it’s a good question
just read all files in directory X in build-time and construct some nice components statically
is what I originally had in mind, via https://nextjs.org/docs/app/guides/mdx
Yeah, figured. That's the simplest approach there is
Thanks, I will whip up a PR tomorrow soon, if you don't mind
we don't post often enough to warrant some fancy system
Is this btw already a step forward to eventually ditch forums?
eventually
👍🏻
Markdown to some components is what we use for our server website, Svelte though iirc.
I'm no web dev at all, but I found that workflow pretty nice and convenient. Writing blog posts locally, commiting to the repo, Github actions builds and pushs it, blog post released. Doesn't need any auth or anything like that this way either
Yeah, exactly why I asked
I'm updating my particle guide for 1.21.9 and need advice. Should I even mention that FLASH supports the alpha channel even though it doesn't have an effect since the particle instantly overrides it?
Might be fixed by Mojang in the future
Just say it does support it but is currently unused
I think that clears all possible misconceptions one might have
Is there anyone who can verify that the information in #592 Add solution for forced hosts not working in velocity faq.md is correct?
I'd love to get it merged, as it seems like a pretty sleek solution to the problem
Well, assuming DNS servers don't do weird stuff that should generally work
lulu, if you'd be so kind as to check out docs#652 for any misinformation (primarily regarding the chunk(holder) info subcommands, I'd appreciate it
Idk if I brought this up before, but an LLM.txt for our docs would be cool
Would be surprised if there wouldn't be a starlight plugin that generates one
What does that do?
Many pages just dump everything into LLMs.txt
There’s a docusaurus plugin

Here ya go
Ew, a spec that wants to add a new "root" special page, instead of it going in
.well-known/
Although, based on
/llms.txt of a website (or, optionally, in a subpath)
Wouldn't that mean you would need to check n URLs for an n segment long URL?
IE
Example.com/en/docs/v1/project/contributing
Needing to check for llms.txt at each level
I don't expect any less from AI bros
Either way they wouldn't check it or ignore it
Just like those asking to not be scrapped
They get ignored lol
@drowsy zinc I did think about documenting the other commands as well, but the other ones which I found were basically self explanatory anyways lol
What command in specific do you think is still missing?
idk, like /restart
the main thing is that if there ever is a command that's not under /paper and you need to document it, you have nowhere to put it without redoing all that
I mean, I would just put it in its own reference/commands category and add the remaining commands there, in case they are note worthy
Otherwise the page also just gets too big if we start putting all possible random commands in there, I kinda like its scale atm
category consisting of The Paper command and The Rest lol
Basically 
no
The initial problem behind the documentation of the paper commands was only because of it containing many developer specific and highly technical/unexplained values
I mean sure, I can change it to general commands documentation, but then people will complain "why isn't /help documented?"
yea it is https://minecraft.wiki/w/Commands/help
Oh well nvm 
Idk apart from /restart there really isn't any command we'd care at all about properly documenting, right?
And I have also not once seen anybody actually use restart before. I believe on Windows it doesn't work at all anyways? Because .sh and what not. Idk the last time I tried it out was probably like 2 years ago
it works on windows but it needs to be configured properly
which no one actually manages to do
Ykw in that case fair enough, that'd require some reading material
- ...I did not know there was a spigot command
- The /tps and /mspt commands, kinda yeah, thought idk to what extend considering the spark variants of those commands are preferred anyways and the spark commands are documented on the spark docs
also the config cleanup is fine by me
I didn't read it all, just gave it a skim that you didn't break anything
I still need to address the one point by lulu (and eternity also brought to my attention that there are config values which do not generate in the config? I kinda wanna document those as well), but good to know
Thank you for looking over that
I mean, the only relevant commands are the ones added by bukkit, spigot, or paper. Minecraft commands are documented in the mc wiki
And there aren't many of those
Idk I will look over it
it's never getting fixed

Classic
why is this happening? The javadocs have been updated?
I see a commit time ago for the stable version
I did
Yes it does?
ah it's under 1.21.9 without the suffix
grr
pull main into your branch for now and I'll fix it later
I mean I did pull it
so that's a new class
rewrite the LATEST_PAPER_RELEASE constant in /src/utils/versions.ts to "1.21.9" for now and we'll revert it when there's a stable build and it can be merged
@drowsy zinc in #1442, the footnotes seem to render really weirdly in starlight. It makes a new ToC entry and also shows the linking symbol when you hover right above it. Imo it looks a bit weird. Anything we (you) can do against that?
no :)
Any progress on this?
I got a general markdown parser down, but apart from that nothing yet. It kinda got lost in my way too long list of projects. Idk how high prio it is, but I doubt it will be finished before like middle october just because of how much is going on in my private life as well atm
You wrote a markdown parser? If so you didn’t have to

Incredible commitment
org ban material
No 
Yes
Whilst writing a markdown parser from scratch sounds like a very me-thing to do, I did not do that
No I mean
It’s already a thing that you can use without writing anything
@pliant canopy @drowsy zinc
also https://vercel.com/templates/next.js/blog-starter-kit (?
but that use just md but its like the same (?
Yes and I am already doing that lol
is it just me or the toggle experimental button is full white?
The problem with everybody being developers is that nobody tests light mode
But, yea, that is busted here too
Cc: @slate scarab
GHOST PINGING
love you
Hi, it's a bit late to ask now, but will help when Mojang does their next release cycle.
I was wondering if the elements in the response from getVersions could include a "version type" like Mojang does to differentiate the Minecraft versions that are release vs snapshot
https://fill.papermc.io/swagger-ui/index.html#/Meta v3/getVersions
such as
{
"version": {
"id": "1.21.9-rc1",
"support": {
"status": "SUPPORTED"
},
For comparison it would be the "type" from https://launchermeta.mojang.com/mc/game/version_manifest_v2.json and I'm thinking would track theirs. I could always cross-query from Mojang's manifest, but it's nice using the Fill API as a one stop place to retrieve latest (release vs snapshot) Paper versions, for example.
I'm bringing this back to life. I'm going to tag @pliant canopy since you originally showed interest in it. I've opened a draft PR that is basic and needs some work done. If you want to work off it, you can. I talked to @opal flare and he likes the POC I've presented in the PR. It'll definitely need some cleanup and some tweaks and actually needs the stuff from the forums migrated over to MDX but I don't think I have the time to commit to do it all myself.
deletion would be pretty bad
information is hard enough to find as is
dont need more dead links
Forum will go read-only
Yeah that's fine, I will check when I am back home if there is anything I've done which I can migrate over and can do the rest
Sounds good!
a little lost... i mean i make works but not sure why works xd
i just change the class for button from text-white to text-black looks like not affect to darkmode
we love tailwind
also if remove the text-black still works xd
maybe its something that got broken with the migration, as in, the old class name didnt exist
well works then i send PR unless i miss something with the tests...
https://github.com/PaperMC/website/pull/157
thanks
I pushed some more changes to the draft pr. It should be better suited to have things migrated over now.
If riley wants to keep the hero images we'll need to backfill / generate some for each post.
ah, they are
would be cool to have filtering by tags and stuff
also not sure I like the blog naming, how do people feel about news?
are the pages pregenerated?
and it would be cool to link the release posts to the versions on the software pages
sorry, just brain dumping, lol
Yes we can make any changes we want. Riley just liked the look so we can do with them or without them. We can definitely add in filtering by tags and have some kind of sidebar. Yes all pages are generated at build time.
And we'll probably have to make some sitemap ajustments but yes we can make it show up there and I was also thinking about adding in a static rss.xml endpoint that is generated at build time as well.
I personally think those slop heros which are shat out by some dum llm are usable for social media only, if at all, but that might be just me
Riley is going to forward some of the photos I sent him here shortly so you can kind of see what we're working with.
oh yes pls
I only added 1 in so that Riley could see it as an idea. We can totally drop them and pretty much do whatever we want.
we can set them as the graph image or something so they show up when being shared, I am just not sure if they look good on a website (I didnt build your branch yet)
I think we can also just use next/og (satori) and have some OG images generated at runtime based on the blog title + tags, etc.
thanks!
thats about as bad as I expeced the hero to look 😛
and the accent color is wrong, nice otherwise
and we can prolly make that release link nicer, I do like the idea too
I'm seeing if I can get the rss route and the sitemap thing in and then you can go ham. There are some slight color changes since those photos since I added in better handling for the backtick content.
So eslint is being removed in NextJS 16. Do we want to go ahead and move over to Biome? I've been using https://www.ultracite.ai/ which is a layer on top of Biome that has been working really well for me.
Alright @spice temple it's all yours if you want it. I updated the next-sitemap and also added in a generated rss.xml route. I'm like 90% sure things will need to be cleaned up, but, it's a base to work off of.
@opal flare Do you know if I have to change this from a draft pr to a ready pr for Cloudflare to build a preview page for it?
I don't want to that this away from you, lol
I just came home from vacation, my to-do list is long enough
I am not sure about biome, it feels like a dead end now that everybody is betting on oxc/oxlint
Going to a hospital on that list?
I told Riley up front that I'm super busy at work and just wanted to get the ball rolling. I'm sure Strok will be taking a look at it as well.
Why does my fucking code formatter need AI? Lmao
don't think it does?
I honestly don't know why it's "AI"
I think that's just tpart of their wanketerring
It's just a set of rules for Biome
"look, we format the code in a manner that works with LLMs"
Idk, I like biome, I just think that it's not what will kill eslint, oxlint will
I may switch out the rss.xml from doing it raw and utilizing something like https://www.npmjs.com/package/feed
Because it's backed by vite/evans company which everybody uses
Probably
Guh okay.
I went ahead and switch the rss.xml route to be generated from the feed package because it ensures it's all organized and clean and stuff. Might need some of the info updated there but it's all in the route.ts file for it.
<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
<title>PaperMC Blog</title>
<link>https://papermc.io</link>
<description>Latest updates from PaperMC</description>
<lastBuildDate>Wed, 01 Oct 2025 16:54:57 GMT</lastBuildDate>
<docs>https://validator.w3.org/feed/docs/rss2.html</docs>
<generator>https://github.com/jpmonette/feed</generator>
<language>en</language>
<copyright>All rights reserved 2025, PaperMC</copyright>
<atom:link href="https://papermc.io/blog/rss.xml" rel="self" type="application/rss+xml"/>
<item>
<title><![CDATA[Paper 1.21.7]]></title>
<link>https://papermc.io/blog/1.21.7</link>
<guid isPermaLink="false">https://papermc.io/blog/1.21.7</guid>
<pubDate>Mon, 30 Jun 2025 00:00:00 GMT</pubDate>
<description><![CDATA[What changed in 1.21.5/6/7, upgrade advice, and acknowledgements.]]></description>
<category>release</category>
<category>1.21.x</category>
<enclosure url="https://papermc.io/blog/1.21.7/hero.png" length="0" type="image/png"/>
</item>
<item>
<title><![CDATA[The future of Paper — Hard fork]]></title>
<link>https://papermc.io/blog/hard-fork</link>
<guid isPermaLink="false">https://papermc.io/blog/hard-fork</guid>
<pubDate>Fri, 13 Dec 2024 00:00:00 GMT</pubDate>
<description><![CDATA[Starting with 1.21.4, Paper hard-forks from Spigot to accelerate development, ship missing API, and simplify contributions—without breaking existing servers.]]></description>
<category>announcement</category>
<category>project</category>
<category>paper</category>
<enclosure url="https://papermc.io/blog/1.21.7/hero.png" length="0" type="image/png"/>
</item>
</channel>
</rss>
also, fyi:
npm warn deprecated @cloudflare/next-on-pages@1.13.16: Please use the OpenNext adapter instead: https://opennext.js.org/cloudflare
Also I did have to update NextJS in this PR for stable turbopack. It wasn't playing nice with next/mdx.
Which means we could try out the next build --turbopack now.
Oh so maybe I do have to manually install it. I thought that was a subset of next/mdx.
Just pushed it to add it back
Hot
Oh
Maybe not hot
I think my browser is just defaulted to dark mode.
Do you want me to shove in OpenNext to this PR?
should be auto-deploying to https://feat-blog.papermc-website.pages.dev/blog now though
separate please
kk
So what I'm going to do is:
- New PR for OpenNext + NextJS 15.5.4
- Rebase blog on the new changes
Okay putting a hold on OpenNext. There's a bug in the current version it seems that is causing issues.
But yeah Blog is free to play around with.
You might have already done a better job than I ever could have made, just basing myself off of the past few messages. My web dev skills are greatly lacking, as I usually only do Java.
Regardless, I can definitely do some tweaks (having solid changes makes it infinitely easier to work with and extend) and migrate over the remaining announcement messages (when I do get home ofc). So yeah thanks for your work, much appreciated 
For sure. Just ping me or something if you have any questions.
I'm sure Mini will continue to rip us apart until it's to his satisfaction 🙂
Mini always does that, learned that when I did my PRs to adventure 😂
Not a bad trait though, asking questions and providing feedback is always good
Haha, it's my job at work
I don't have the luxury to write code myself most of the time anymore
I believe people call that suffering from success
I got to rip apart two weeks of progress at work today, that was fun ^^
Only two things I would have done differently tho and those require a bit of work, so don't even blame my team
My PO had so much anxiety of me being gone for over two weeks, as if my team was a bunch of idiots that need me to tell them what to do
If you’re not the PO what job role do you have lol
CBO
Officially? Just a normal team member
What best would describe it would be a lead dev
PO is business side
Ah sure, makes sense then
@drowsy zinc is there a reason why you removed the automatic adventure versioning logic in the adventure docs merge PR?
the current site isn't automatically versioned
I don't think it is bad though?
Maybe they just didn't think of it or wasn't possible with the old tooling, who knows
I figured zml and co. preferred updating it first and then bumping the references
You figured or you asked? Just wanna make sure
Imma ask just to be certain. @timid cedar do you want to manually bump the dependency versions on the docs when you do an update (just editing like four values in a file), or want it to be automatically fetched on build time?
figured
Either way we'd have to manually trigger a build right? Idm then
Basically the same amount of work either way
it's not going to be me (or you strokkur) updating it most of the time so I tried to adhere to the current workflow
Well whatever, will keep it as it is. It can always be automated at a later time if you do decide to want to have it automated
It would be nice if the builds themselves were automated, but idk how/if cross-org github action triggers are even possible
we manually trigger bumps for paper too so
Apparently it is possible: https://medium.com/hostspaceng/triggering-workflows-in-another-repository-with-github-actions-4f581f8e0ceb
Is this something we are interested in taking a look into and should I spend time investigating it, or not?
I don't think it's worth the complexity but I don't mind having it
Alright, I'd love to get the adventure docs migration finally merged. I went ahead to do some final touches. (#573: feat: Add Adventure docs). @drowsy zinc @timid cedar if you could go over it quickly, that would be really great.
Giving it a good look over now! 🫡
Well, "quickly" going over 4k lines of changes definitely ambitious, but content wise I am pretty positive that it is 1:1 with the current kyori docs, so mostly style and feel things are what should be checked
Are the URLs the same? e.g. can we just rewrite the old url to the new one easily?
I believe the only difference is that there is no .html suffix and I also think the index.html ones are omitted
Apart from that, I think so? Shouldn't be hard to write up something simple to redirect it
on the pr
PR is better
Alright, I've tended to the review comments (thank you kezz!). Anything else we want to do before I merge this (I'd like an explicit yes from both of you @drowsy zinc and @neon epoch before that, as this is a big PR). Minor stuff can always be done afterwards as well (like inline code highlighting), so it is not the end of the world if something is found afterwards.
-# Also sorry for the many pings scorp, I hope you forgive me 
Yeah I can give it a look over now, I’m stuck in a useless work call
I was very confused why stuff wasn’t working until i saw I was logged in with my work account 🫠
1 change for the releases that can come down the line, couple of small issues - impressive work!
Thanks for the review, I changed everything you requested to

But yeah, I think it’s good and I couldn’t see any glaring issues - nice work
is it good to merge now? @drowsy zinc
because if so, can someone make a mapping of old -> new url? and I will apply redirects tonight
https://docs.advntr.dev/getting-started.html
https://docs.advntr.dev/community-libraries.html
https://docs.advntr.dev/faq.html
https://docs.advntr.dev/audiences.html
https://docs.advntr.dev/text.html
https://docs.advntr.dev/serializer/index.html
https://docs.advntr.dev/serializer/json.html
https://docs.advntr.dev/serializer/gson.html
https://docs.advntr.dev/serializer/legacy.html
https://docs.advntr.dev/serializer/plain.html
https://docs.advntr.dev/minimessage/index.html
https://docs.advntr.dev/minimessage/format.html
https://docs.advntr.dev/minimessage/api.html
https://docs.advntr.dev/minimessage/dynamic-replacements.html
https://docs.advntr.dev/minimessage/translator.html
https://docs.advntr.dev/serializer/ansi.html
https://docs.advntr.dev/bossbar.html
https://docs.advntr.dev/sound.html
https://docs.advntr.dev/title.html
https://docs.advntr.dev/book.html
https://docs.advntr.dev/tablist.html
https://docs.advntr.dev/resource-pack.html
https://docs.advntr.dev/localization.html
https://docs.advntr.dev/platform/index.html
https://docs.advntr.dev/platform/native.html
https://docs.advntr.dev/platform/bukkit.html
https://docs.advntr.dev/platform/bungeecord.html
https://docs.advntr.dev/platform/spongeapi.html
https://docs.advntr.dev/platform/modded.html
https://docs.advntr.dev/platform/fabric.html
https://docs.advntr.dev/platform/neoforge.html
https://docs.advntr.dev/platform/viaversion.html
https://docs.advntr.dev/platform/implementing.html
https://docs.advntr.dev/migration/index.html
https://docs.advntr.dev/migration/bungeecord-chat-api.html
https://docs.advntr.dev/migration/text-3.x.html
https://docs.advntr.dev/contributing.html
https://docs.advntr.dev/version-history/index.html
https://docs.advntr.dev/version-history/adventure.html
https://docs.advntr.dev/version-history/adventure-platform.html
https://docs.advntr.dev/version-history/adventure-platform-fabric.html
it's identical but swap the main bit and remove .html
ok, I will write a rewrite rule when I get home unless someone makes one for me
You just need a cloudflare redirect rule?
Yes. One for root without .html and one for all .html
Alright, I will play around in my own cf and then send you something soon™
Thanks
Alright, the following should work:
- Redirect Rule: Wildcard pattern. Request URL:
https://docs.adventr.dev/*.html. Target URLhttps://docs.papermc.io/adventure/${1} - Another one with the same target url, but without the
.htmlin the request URL.
Make sure to also tick the preserve query string option.
I tested it on my subdomain adventure.strokkur.net to point to the preview deployment, and seems to work fine: https://adventure.strokkur.net/migration/text-3.x#append-with-a-string-argument
should be
Hey @pliant canopy I might have some time today / this weekend to do more stuff with the blog. Have you had a chance to look at anything yet? Probably need to work with Riley to go over the changes that need to be done and then just migrate things over.
Also @opal flare to prepare for OpenNext, we need to drop next-sitemap and use what NextJS has integrated already. This is because the postbuild doesn't run properly with OpenNext so we will have to build the site once and then run the OpenNext stuff. Do you want a dedicated PR for migrating from the next-sitemap over to what NextJS provides now?
I already have it done fwiw. Just don't know if I can shove it in blog branch or you want a dedicated PR and go ahead and get it in.
Edit: Going to just open a dedicated PR for it.
The new thing that is replacing next-on-pages.
Working on too many branches and keep copying the wrong things over 🙃
I removed the blog stuff from that PR. I'll just have it marked locally to add into the blog branch when this gets merged.
I got back yesterday, but I feel horrible right now, so I haven't had the chance sorry
Absolutely no worries. Focus on yourself first 🙂
Yeah I am gonna take a nice bath now and then gonna to sleep
Eventhough it is just 8pm
Combined PR for both is fine
@drowsy zinc merge when ready and I will deploy redirects in a bit
Just ping me when merged
Able to give it as a Cloudflare rule string? :p
Ope. I already opened a dedicated PR for the sitemap change but I can rebase the open next branch off that.
is there a reason why its a normal merge and no squash?
need to (partially) revert a bug workaround commit later
Sorry I went to bed. Do you still need it?
(how do you do that anyways xd)
Edit the expression and copy paste, and also give me a dynamic expression for the second
Yay!
You already got it, but just for fullness, here is what I had:
-
(http.request.full_uri wildcard r"https://docs.advntr.dev/*.html") -
wildcard_replace(http.request.full_uri, r"https://docs.advntr.dev/*.html", r" https://docs.papermc.io/adventure/${1}") -
(http.request.full_uri wildcard r"https://docs.advntr.dev/*") -
wildcard_replace(http.request.full_uri, r"https://docs.advntr.dev/*", r" https://docs.papermc.io/adventure/${1}")
advtr.dev 😭
Horrible domain to type out by heart
an issue of skill
it's a lot easier to remember when you know that it's adventure without the vowels
that makes a lot of sense, always thought it was just a weird way to abbreviate adventure
but with the a
dvntr
With this then make sense change the url into the adventure repo?
also i notice the discord mentined its still the one with the channels related to adventure are closed in favor of #adventure-contrib
Yes I'll do a pass updating everything and archiving the old docs repo on Monday
https://github.com/PaperMC/website/pull/161 this is ready for merge btw. Just needs approval.
the old adventure docs repo has a few issues i'd like to migrate over, would it be worth having a tag for the project type perhaps?
Yeah, for sure. Feel free to migrate them over and I will apply the tags
okie dokie, heads up for the issue spam :p
misunderstands
I am an English
should've just added a [sic] to every one of @nimble marlin's issues yeah :p
hey
👀
was?
or im see the wrong README/about xd
ahh i was talking about the README/about of https://github.com/KyoriPowered/adventure/
that has the other discord and docs urls
and notice mentions in class
i can open a PR for that things (the about not) or just tell here?
Oh right I see - I'll just update that now

more like sensory overload
merging now
https://docs.papermc.io/adventure/platform/modded/ historic versions here is pretty outdated if anyone feels like updating it :D
about to push 6.7.0 for 1.21.9
any chance for a little review in https://github.com/PaperMC/website/pull/155 ?
if its valid the fix
ready unless bad read the releases page xd
I lov
e css
If anyone is good with colours: https://github.com/PaperMC/website/pull/163
help
@viscid thistle ^
What page is this on
PaperMC is a Minecraft software organization focusing on improving the game's ecosystem with faster and more secure software.
not?
https://feat-colors-papermc-website.papermc.workers.dev/downloads/paper
- used for the "1.21.8" colour, the download button, and the other build listings below
It needs to work on both light + dark themes, and the text needs to be visible in all cases
in dark mode feels.. too.. palid?
Colours can be completely changed, but the general theme is:
alpha: red
beta: yellow/orange
stable: blue
recommended: green
so as long as they follow that theme, any changes to the hex colours used are fine
any recommended currently?
easiest way to preview is to use it locally and chage the colour for another channel 
only old builds of velocity are recommended channel
i found one xd
yup
The colours also need to "match" each other - similar kind of "tone" or whatever. Per-theme (light/dark), at least
For the blue:
Dark:
- bg #4DC4FF
- text #0A192F
- hover #37B7FF
Light:
- bg #007BE0
- text #EAF7FF
- hover #006BC7
You can push to branch if you want
I have games open
Light:
alpha: { bg: '#FEE2E2', text: '#B91C1C', hover: '#FCA5A5' }
beta: { bg: '#FEF3C7', text: '#B45309', hover: '#FCD34D' }
stable: { bg: '#DBEAFE', text: '#1D4ED8', hover: '#93C5FD' }
recommended: { bg: '#DCFCE7', text: '#15803D', hover: '#86EFAC' }
Dark:
alpha: { bg: '#3F0E0E', text: '#FCA5A5', hover: '#DC2626' }
beta: { bg: '#422006', text: '#FCD34D', hover: '#F59E0B' }
stable: { bg: '#0D1B2A', text: '#93C5FD', hover: '#3B82F6' }
recommended: { bg: '#0F2E1E', text: '#86EFAC', hover: '#22C55E' }
Try that
@layer theme {
@theme {
--color-alpha: #FEE2E2;
--color-on-alpha: #B91C1C;
--color-alpha-strong: #FCA5A5;
--color-beta: #FEF3C7;
--color-on-beta: #B45309;
--color-beta-strong: #FCD34D;
--color-stable: #DBEAFE;
--color-on-stable: #1D4ED8;
--color-stable-strong: #93C5FD;
--color-recommended: #DCFCE7;
--color-on-recommended: #15803D;
--color-recommended-strong: #86EFAC;
}
@theme .dark {
--color-alpha: #3F0E0E;
--color-on-alpha: #FCA5A5;
--color-alpha-strong: #DC2626;
--color-beta: #422006;
--color-on-beta: #FCD34D;
--color-beta-strong: #F59E0B;
--color-stable: #0D1B2A;
--color-on-stable: #93C5FD;
--color-stable-strong: #3B82F6;
--color-recommended: #0F2E1E;
--color-on-recommended: #86EFAC;
--color-recommended-strong: #22C55E;
}
}
Might be closer
that format is still invalid - that makes like bg-channel-alpha-bg
4
That might be better
Multitasking atm
<button class="rounded px-3 py-1.5 font-medium
bg-stable text-on-stable hover:bg-stable-strong">
Stable Build
</button>
I'm a good namer
ok, let me frankenstein this
🫡
because there are differences too
hmm
yeah I think someone else will need to do this
I am dumb
You could work on the ISR cache if you want something to do 🙂
The colour changes are required for other moving parts; a blocker :p
Not sure I'm a fan of these
too dark for sure
@undone drift 
Can you create an R2 bucket and give me a name and I'll push something for it
for cache?
Ye
It looks like that's what we have to do since we switched from pages to workers to get ISR back
website-cache
Why did the deploy instantly fail lol
text: 'Could not resolve "@opennextjs/cloudflare/dist/api/overrides/incremental-cache/r2-incremental-cache"'
gotta love that
it is
Idk how it's different then whut
Oh I got it, I think.
It was right but our tsconfig.json was still from our Next 13 days I believe.
Okay it deployed but it's still saying miss
Wonder if it's cause preview?
Is it tied just to production?
Might be. Anyways, I gotta 😴
updated colours
That isr stuff broke the download page for paper, it throws a 500
Deploying an old build
No clue where the fuck cloudflare hides the logs for this
Deploying the old build fixed it https://papermc.io/downloads/paper
Looks to be broken again... something is up 
yeah I dont know what's wrong.
That's indeed frustrating
it worked for a minute before, wtf
it did work for me when Mini said it was fixed.
then it broke again when I check shortly when I saw a report in #paper-help
I'll get to my PC in a bit and just disable isg somehow
restored for now
What did you do?
can confirm that
reverted to an older deployment
That's what I did too 
probably mobile being dumb
Let's see if it breaks again
if you have time to look btw @spice temple
i can try using that blue for the stable channel i suppose
thats the only thing I noticed, other than our light theme being really meh to my eyes, but my room is also kinda dark since the sun isnt up yet, lol
I think I prefer the darker one, but generally, just keeping the color palette limited (and I think the dark blue is our brand color?) is a good idea

