#docs-website
1 messages Β· Page 15 of 1
in general not sure if we even need them all, but they look good
Hello!
good enough to merge for now?
I presume it is know here that I can't access "https://papermc.io/downloads/paper" as yall are working on it right now? Just wanted to verify it wasn't an issue on my end
Such a real one, Thank you so much, riley. Decided to do some updating tonight and noticed the issue 
Back in 3, checking your changes then and approving
too late, i used it to fix prod 
I blame the cashier for being slow
and here is where you tell us you used the self-checkout

hm, broke again
found the issue
nah self checkout is dum slower because I need to do double the work
kaufland k-scan π
Props to the new adventure docs looks very nice
Site loads so fast now. Nice work on the quick fix.
only me feel "strange" for that ugly embed message? xd
really fast just still that issue for velocity downloads
Well, that's just what it looks like
yeah... the website not run like the docs site in terms of cloudflare?
because docs the preview thing looks different
The docs and the website have entirely different cloudflare/github actions setups
We also had to migrate the website from Pages -> Workers yesterday.
Server Side rendering?
pages can also do ssr but cloudflare is moving away from pages to workers with static assets
So next-on-pages has been fully deprecated and they are having people move over to OpenNext. Part of it makes sense, part of it seems like they want to try to make a little more money.
So we're using OpenNext with ISR via queues and r2.
I saw in Nuxt etc pages can't be ssr because just staic rendering
We initially didn't add in the r2 and queue but I had noticed that things felt slightly delayed when we originally pushed this and then I realized it was never doing a cache hit for ISR (main page, downloads, etc). So that got added in last night and what you're seeing on the page is actually just a cached entry from a queue / r2 bucket cached on the edge.
All pages just seem... instant now. I love it π
And then when the blog branch is ready to go, we'll be upgrading from NextJS 15.3.5 -> 15.5.4. Shouldn't notice anything on the frontend.
idk I don't see anything that says nuxt ssr is not supported
pages has pages functions for server-side code
yay!
is strokk going to have to write docs for it ;)
The docs for that already exist: https://docs.papermc.io/misc/downloads-api/

should rename that to Downloads Service too
Fair enough
you want to PR that one?
I am about to just commit
smh
Small and simple change, a few words changed, and a redirect added
There; fun and done
oops I cooked a link
ty
the redirect won't work for links with a trailing slash (any link pasted out of starlight)

it didn't really need one imo
the slug change wasn't needed
That I can get behind
does something have to be done for the dl page to show .10
Have we had a situation where we had another version go experimental before the previous went stable? Maybe it's being confused on 2 different experimentals? Not sure.
I see the 1.21.10 in the build explorer but idk what controls what it shows as the experimental build version toggle.
Paging @opal flare
Did it not need a redeploy?
I don't think so? I see the API calls for the 1.21.10
I guess I can go look at how it determines what to show.
re-deploying
I know we've had a new experimental drop before another version was done, fairly recently too
.10 now live
What's being cached that requires a re-deploy for it?
no clue
.5 .6 .7, yes
it's apparently just all statically done or something
Going to investigate this.
Why do we have a fill and a service lib? What's the difference? Once has bstats, one doesn't?
I can't even see the downloads.ts context being used anymore.
Honestly, I think somebody submitted a PR which moved stuff over and we hit the button and never looked at it again
Time to start rolling a dice and removing stuff
Ope we just rolled a 6 on downloads/all/page.tsx! It's gone!
Is it just me or am I the only one that noticed that in the past month(s) this channel has become quite active
We bullied some people into forming a docs team
We keep telling them, 100 more pages, they can have the keys
it is also one of the first channels hit by spam bots so people are always like "what did u say m8?"
(I've never seen anybody do that)
what did u say m8?
I was channeling Australian from @astral smelt
Same difference
but yes, this channel is basically "all web properties" these days
website, docs, fill, fill-ui
web is the future
I hope not
2030 paper in the browser
I hate web
Do we really need to SWR update the bstats on the home page? Do we expect people to sit there and watch it?
If not, I'm switching that over to ISR.
Yes
We have unlimited Cloudflare Workers, right?
This team page is 25MB because of all the images lol. I was going to ISR it and optimize it.
we do
Bet. We're going to use that.
I think I also solved the redeploy issue. TLDR the route gets statically pre-built and even in ISR it just re-uses the fetches from the download.ts so I added in revalidate instructions to it.
oh good
Alright going to open a PR for this. I think it's a good first iteration of cleanup.
I'll save the migration of eslint or biome for another branch.
sure
Let's pull this up on a preview branch just to ensure things work as expected. Let me get the PR open, one sec.
just out of curiosity... what should the redirect be so that it redirect correct? It's formatted the same as the other redirects, as far as I can tell
it needs one with a trailing slash
other redirects are from docusaurus-era which had a different file structure, so they didn't have the trailing slashes
Thanks Cloudflare. Where's the URL? Smh.
I eated it
okay then
wait, so just one with a trailing, or one with and one without?
/misc/downloads-api /misc/downloads-service/
/misc/downloads-api/ /misc/downloads-service/
What if somebody manually removes the trailing slash because they feel funky?
@opal flare Give pls π
Mhmm
both
If it's the same general URL scheme for branches then I can at least start guessing them.
it does not have a preview for some reason
o.o
ayo that's cool
minus hangar, that one has it's own channel so that it can die alone

Well in happier news that 25MB is down to 7MB now
Only works for one day ._.
The ISR setup that we had to go with on the migration from Pages to Workers disables previews from working.
It worked yesterday because I didn't fully setup the ISR properly 
Then now its impossible has previews in PR?
It's not impossible but we'd have to do automation via GitHub Actions to spin up and spin down environemnts per-pr.
Time to bring back the Astro rewrite at this point


0 days since last migration
Next.js is just such a headache
What it boils down to is that if we're moving the build explorer to fill and off the main site, we have minimal need for any JS on the site.
This exact sentence. I said it 1h after using Next. Never touched it again.
Right, I saw that message, but that's just EterNity, nothing official
Why not plain html? Fully optimized to the teeth
Because Astro is the good in-between. Markdown support for blog, etc
I am joking here btw
I think we could push through an Astro rewrite if we do push it through
But I think having a clear decision on that would be great
Do we want to rewrite the site to Astro definitely?
I am 100% open to the Astro rewrite. Using Next.js has always been a headache:
- self-hosting is not really possible in a sane way unless you jump through hoops, and has no previews
- cloudflare workers hosting works but requires an R2 bucket + Durable Object, and has no previews
- astro lets us use markdown, react/svelte if we want to, and is already in use for docs.papermc.io
Let me see what I can pull from the old branch and see what it'll take to update it.
Whilst we are at it, any functional differences we'd like to the site?
Might as well get some plans in if we are touching basically everything already anyways
blog posts or similar
Ideally we do a 1:1 copy first and do any changes after that.
Fair enough
We tried and generally failed due to the interactivety stuff being cludgy and people generally being against typical frameworks used to solve this
The biggest blocker was the build explorer... but if that's confirmed being moved to Fill...
The biggest blocker was the downloads page for paper, etc, itself
@opal flare?
I haven't heard anything in that regard
Only EterNity making things up
does it have to be that fancy?
just markdown files + parsing on redeploy would be enough for the blog post once per update
it was stuff like having multiple artifacts for something required some interactivety that was basically unviable without a framework or raw JS
I guess that's not a big issue now as we don't have multiple artifacts on there
framework of choice is svelte these days - diffs.dev, fill-ui, and others use it
which is lightweight too
I think svelte was loosely mentioned but nobody had any experience with it at the time
Yeah I haven't really used it much.
its pretty easy to learn if you know others
svelte and mdsvex for the markdown parsing is what we do for our server website for the blog posts there.
The workflow is for posting something nice and easy imo. Though a friend made the site, idk how svelte was to work with
The benefit of Astro would be the blog section. It just... works. It's designed for it.
- all the images created at build-time.
I mean, the issue was that our downloads page had a whole bunch of drop downs and astro had 0 means of being able to cleanly have standard dropdown menu stuff without running around to more legacy web hacks
and the options were basically use something we already knew, which was blocked; write JS, which sounded tragic; or pick up svelte, which sounded like a task for next year
we also use svelte in the docs
for the item command converter and start script generator
yup
Would it be raw Svelte or SvelteKit?
sveltekit is what is used iirc
What portions of SvelteKit vs Svelte would we need for interaction on the download pages?
Because we could just shove Svelte in those pages with Astro so we can maintain the build-time responsive images.
sveltekit is the routing+ssr+other stuff, svelte is the actual framework with the markup and reactivity
Correct
I'm down for whatever. It just sucks that the integration for Cloudflare Workers with NextJS causes so much limitation.
Astro + Svelte
Because I have to ask: Are we sure these various services will still exist in six months in a way that is useful to continue using them? Because so much of this js stuff comes and goes like an indecisive toddler.
Astro is not small; used by companies like Google, Microsoft, Cloudflare. Sponsored by Cloudflare, Google Firebase, Netlify.
Svelte is owned and backed by Vercel.
vercel, evil
jokes on you, Nuxt is also owned by vercel
All 3 major frameworks are owned by them lol
vue too?
No, think that is still independent to some degree
Vercel owns Svelte/SvelteKit, Nextjs and Nuxt
so vue ππ as always
but vue is indeed nice, fell in love with it (coming from svelte)
One of the things I like the most about vue compared to svelte is:
{#if porridge.temperature > 100}
<p>too hot!</p>
{:else if 80 > porridge.temperature}
<p>too cold!</p>
{:else}
<p>just right!</p>
{/if}
becoming
<p v-if="porridge.temperature > 100">too hot!</p>
<p v-else-if="porridge.temperature < 80">too cold!</p>
<p v-else>just right!</p>
it's chef's kiss, I was tired of the svelte syntax for ifs/loops/etc
love the rest, but this one point was annoying
NuxtLabs is. Not Nuxt.
Sure, but NuxtLabs is literally the maintainers of Nuxt
That is true, but, there's a key difference of owning the framework vs just hiring a group of people that work on it.
But, neither of those will be our problem for much longer.
How far along is oxlint nowadays? I feel like that's the future
Fwiw Vue obviously has easy markdown blog support (nuxt content)
It's not, svelte is independent, just part of the core team is currently employed by vercel to work on svelte(kit)
Same here
Only next is "owned" by vercel iirc
No, not all
Before anybody wastes time on Astro again I would like to see a short diagram or description on how the components are fractured, seeing as you can't reuse server rendered components on the client side dynamically, because astro can't be used client side
And you can't use client dynamic components on the server
(I literally wrote my own "astro but one framework for both client and server" meta framework at work because of that shortcoming, once nitro 5 hits am looking to refactor it into a vite plugin and then pressuring people to open source it)
@spice temple please review
Is performance actually an issue?
I'll look at it after this meeting
These mongo index thingies seem like dum API, lol
But looks good
One day I'll push some integration tests to make sure these changes don't break @steady fulcrum 's stuff, lol
Honestly no idea. Never used it.
I switched from eslint to Biome
most people I know (which don't use eslint/prettier) use oxlint for linting + biome for formatting
Interesting
biome can also be used for linting
I know that biome did a big push to pass the prettier test suite, I guess the oxlint formatting isn't that far yet
proper mdx support is lacking in all of them sadly
Yeah, but in the future, when we all use vite with rolldown, there are giant speed gains if all your tooling uses oxc under the hood
Bundling, linting and formatting all in the same ast, not multiple parsing steps
Are you talking about in terms of using Astro for the site itself in relation to Paper?
I would believe so
to be honest, the fact that we're noy relying on serving multiple artifacts for a single thing makes it simpler, idk if we want to keep support for that or not though
That was a huge contention point, as we couldn't figure out how we were going to implement that without a lot of headache
I think we need firm confirmation from Riley on if the build explorer is being extracted and we're just linking to Fill. That would provide some insight on any challenges we might face. Because if we're extracting that, then we can just use a Svelte component for the Terminal and the downloads page for showing the latest builds + experimental. The only other thing I can think of would be bStats which we could also turn into a Svelte component and just fetch the data or do it in raw JS.
The blog portion should be super simple due to Content Collections.
Not talking about the explorer
Download the latest builds of Paper, the high-performance Minecraft server software.
That page used to support multiple artifacts being published for a single download
Ahh yeah okay that will be something important to discuss too. We could do fetch with swr in a Svelte component most likely for each of the download pages, but everything else can be static.
Riley said CORS shouldn't be an issue, so I think we should be able to fetch builds from the API directly in the component.
For that we had stuff like dropdown menus which was a huge issue because to handle that you either need to pull off some hacks like relying on eating up the entire page with a transparent div to support closing the menu, or figuring out some way to deal with that state
Yeah I gotcha
So, core team members of svelte/nuxt are employed by vercel but vercel doesn't own it?
I think it was basically one of those "we pay them to do stuff but we don't influence the project"
at that point why pay them tbh
Maintain the ecosystem, especially when you've spend your own company hours into making their integration into your platform almost a 1 click deploy
good point
is the custom font key valid in this case? AFAIK it should be in quotations
can be quoted or not quoted, doesn't matter
It's just two args
I just mean that we should have a plan because else countless people will waste time again when we run into the same challenges again
Not all core team people are employed by vercel
When does Paper 1.21.9 come?
Wrong channel also see below @frosty shoal
Yes
π
It's available already on the website
Beta
Just listing out some things for me to attend to later for Astro branch:
- Utilized the new responsive image tooling Astro provides
- Utilize Svelte for download pages and Terminal
Maybe SWR for the download page in Svelte for fetching buildsSWR is React only. Could just use localStorage as a cache, idk? 100IQ.- Lift and shift the TailwindCSS config
- Add (separate branch?) for starting on content collections / blog
- Ensure sitemap and robots work
- Cloudflare Pages integration
Am I missing anything here for high-level notes?
I've started working on this locally and I'll push what I get done tomorrow.
I just pushed some more changes (https://github.com/PaperMC/fill/pull/8/files) - another look over would be good
went from ~6000ms to ~273ms for one endpoint (when uncached)
So, just so I understand, you are moving the website/etc to Astro?
Website, yes
Docs already use Astro, fill/build browser wonβt use astro
Yeah just the website. The Cloudflare Pages adapter for Next went deprecated to we had to move it over to workers. To get ISR working with Next + Workers, we had to add in R2 + DO adapter and that worked, but, when you add in a DO to a worker you can no longer easily have PR branch previews so we're basically SOL unless we migrate off Next.
Hey, I celebrate whoever migrates off next xD
I'm fighting some styling stuff, but, I think I might leave that to someone else.
Alright. Going to push what I have. It's still messy, but it's a start. I have to start packing for a weekend trip. Someone who knows Svelte better than me please look at the CLS issues on the download page. Javadocs page is missing, some utils may be missing, needs a bit of TLC for styling.
the layout shift is probably because it's marked client:only
should be client:load
I'm not seeing a difference in the CLS between those two, but, I'll just have someone who understands it better take a looksie π
Is it rebased off main?
Negative. I just pushed new to the astro branch to ensure all the Next stuff was out of it.
ah okay
I can work on rebasing it here in a sec.
Or maybe not, it's giving me a lot of complaints. It's lil bit of a mess because the old Astro was on pnpm so it has to migrate the pnpm -> bun and then the prettier / eslint.
Alrighty I'd say it's at least 50% done. Styling tweaks and the CLS stuff and once that's done we can add in whatever linter / formatter and then blog and call it good.
woo!
graphic design is my passion
They should get a mention somewhere on the doc site once CurseForge version is approved right? 
Yes I think so
I think currently riley is coping with the curseforge signup system
(it's not good)
at this point i don't know if i want to
The Astro branch I think is ready for performance and UI review. Once that's good we can decide what linter / formatter to add in, work on blogs, and then get this live.
Someone who knows Svelte better than me should probably took a look at the components, but, they work.
It also needs a rebase on main so good luck to whoever handles that π
I volunteer @drowsy zinc
Tbh I would just do a git branch -f main without rebase
And then force push main 
I'll let Riley make that decision 
I know that when I looked at it last time it was in svelte 4 reactivity syntax and didn't use runes
It's using Svelte 5.39.11 atm.
I think there's an issue in the terminal with the way the lines show up so that might need to be looked at as well, but, for not really touching Svelte before, I think things turned out somewhat decent.
why don't I see this branch th
feat/astro
ah now it appears? lol
GitHub thingsβ’οΈ
yeah but the components use legacy svelte 4 syntax, could probably run sv migrate to update it
yeah noticed that too
while it will still run in the Sv5 engine, it's always better to use the new syntax
Also, for SoftwareBuilds.svelte, any reason you didn't make project and version props required?
And, maybe nitpicking but you are creating the same Date object twice on line 61 and 63
In SoftwareDownload.svelte, not sure about $: version, id, fetchBuilds();, pretty sure this is incorrect svelte syntax and not how reactivity should be done
for the correct syntax you'd replace that with an if statment and then call fetchBuilds()
and if you do that, you don't need onMount(fetchBuilds);
In SoftwareDownloadButton, I'd probably encase navigator.clipboard.writeText(d.url); in a try/catch. If I recall, some browser's private/incognito modes might break this and it would throw errors.
Overall seems pretty solid though
Feel free to push a commit to clean it up! I don't know much about the Svelte 5 syntax, so, any improvements are more than welcome π
I am heading out rn sooo, but if @drowsy zinc has free time, feel free π
Bold of you to assume he has push perms xd
I'm gonna put all that text into claude and let it rip
π
Fork and push a PR to the branch and I'll merge it in
I actually have no idea how claude 4+ is with svelte
@alpine storm if you ever do want to do a conmit, make it and send it to me with git format-patch and I can push the commit for you
fuck Discord
@lean venture
I switched to vue/nuxt before claude 4+ came out
our dear soldier scorp will handle the small changes and migration to svelte 5 π«‘

now, if ya don't mind, I am about to go eat something and get drunk
have fun
I just need to know if anyone is serious about working on the Svelte stuff cause I'm going to work on some misc style stuff that I can find and I don't want to butt heads with anyone π
check in with scorp, I just sent a review of the svelte stuff from what I could see
but overall tbh it looks perfectly fine
I was able to run the svelte 5 migration on it
Oh and the waterfall download page I think needs some support for matching the EOL messages.
I'm going to push these changes for the svelte 5 migration
Oh I think I got the Terminal output fixed. Going to push this too.
Fixed this
Okay, pushed a fix for the waterfall download page message, also fixed the missing colors for the waterfall download buttons, I'm probably forgetting something. I think it's about ready for review? Still needs to be rebased and we need to decide on oxlint or biome and then look at content collections for blog.
Glare is on a streak
I'm seeing if I can attempt this disgusting rebase.
I'm on commit 61/68
Got it done, just doing some fixups.
https://github.com/PaperMC/website/pull/166
Alright, it's ready for review. Couple things:
- Re-do the readme and make it pretty? Idk
- Decide if we want to add in oxlint / biome before we merge this in or not
- Decide if we want blogs in before we merge this or not (I'd say after so we can at least get this initial migration out)

Yup, it's a big one due to this getting rebased so it went through pnpm -> bun and then biome was removed for now and then migration of everything react over to either svelte or astro.
- maybe because its the default astro readme no? pretty sure not need the
Astro Starter Kit: Basics how to create a project - maybe in another? i think this whole PR can be a "squash and commit"
the only way to preview all this is well checkout and run in local not?
Correct because we have previews Disabled
It'll go -> confirm things locally -> merge it in -> migrate back from workers to pages in Cloudflare
this might be my new hut since #hangar-contrib kinda died
Working on an updated README now.
How about that?
Ignore the branch I left that in on accident for the clone command
I need one of those pretty readme generator tools
migrate back why
We migrated from Pages -> Workers a few days ago and that's what ended up making us want to drop NextJS again.
We can't do preview builds with the way we're handling it on workers right now.
So this Astro migration will allow us to go back to pages.
Oh we also need to add in the Cloudflare adapter once we confirm everything looks good.
Actually I'll just add that in now.
Just need someone to do the README and we need to decide on the linter / formatter and then it's ready for testing. Would be cool to see this get merged by this weekend and we can start working on the blog.
Because 99% of the site is static data and we were originally on Vercel but they wanted an arm and a leg after changing how they handled OSS sponsorships.
totally not biased but ever thought about hosting it on Appwrite Sites xD
I could potentially organize a sponsorship
So much of our stuff is already in Cloudflare and it blends the analytics.
We generally don't want to rely too much on the generocity of 3rd parties, especially smaller entities
We've generally just been bit by it far too many times; being vendor agnostic would be a good idea, being able to spin everything up on a VPS tomorrow would be nice
(also FY it's self hostable)
So that took us over to Cloudflare Pages utilizing next-on-pages for NextJS. Then next-on-pages recently went deprecated and they started suggesting a project called OpenNext, and we got that done, but that required us to migrate the site from Cloudflare Pages -> Cloudflare Workers. After getting ISR setup with NextJS on OpenNext, we found out that there's really no easy way to allow PRs to create preview envrionments without having to hack GitHub Actions. It also brought to rise a pretty big security issue with the default setup for how it all worked.
Fast forward to now, we're migrating to Astro, switching back to Cloudflare Pages, which then we will be able to do preview branches like normal.
yeah understandable
was just throwing it out as I know that appwrite has sponsored several OSS projects before
and well, can be self hosted anyways too
but yeah, CF Pages is awesome either way
and if it works, it works
We originally tried last year to do Astro but at that time we didn't have Fill v3 and it's UI. That was the major blocker for the Astro migration because instead of having the build explorer in Astro and trying to hack in reactivity for that, we just link to Fill. Which that reminds me, I need to add in a link to that on the downloads page.
Yeah makes sense
as said, was just pointing it out, I'd rewrite hangar using appwrite if mini allowed me but I understand the whole vendor lock-in, even if it is self hostable
looks fine maybe a badge for status and other things like the docs repo?
build explorer links to the proper fill-ui URL now based on the project being looked at.
I gotchu. Gimmie a few.
-# that space is for ads?

i forget the - sorry xd
hmm the node min version?
It's Bun
ahh okay.
the deployment status not? if that works like the same than docs not?
We might have to wait until we deploy on Cloudflare Pages to get that
I'm so good at READMEs 
Going to just roll with this and someone else can touch it.
Need to get back to work. Good talk people. Ping if needed.
Itβll stay on Workers anyways, Pages is deprecated it seems
so previews still won't work?
no
you enable it in the wrangler config
if you mean for outside PRs it's the same situation as pages where you need a third party action afaik
I don't think it's officially deprecated, they just suggest Workers for new projects
I think the issue we had with Workers for previews was the ISR in OpenNext required a DurableObject and Queue cache.
But I think previews work normally without them.
Yes
i notice a the docs still mentions docs.advntr.dev im going to open a PR for that but also
https://docs.advntr.dev/minimessage/ throws time out and not redirect to https://docs.papermc.io/adventure/minimessage/
-# im bored in the meet and goes to the docs for.. dont remember why and find this xd
bad jokes = org ban
You clearly haven't seen my past messages in the other channels
I pretend I do not see it
When I have nothing to do I just start spouting nonsense
weird, https://docs.advntr.dev/minimessage/ works for me
same
interestingly without the trailing slash it just redirects to the root adventure docs
a timeout would mean cloudflare is dead, since docs.advntr.dev is a cloudflare worker
this is an error in my redirect logic, I will fix in a few
now works... region issue maybe...
that would be my guess
smh
https://github.com/PaperMC/paperweight/blob/main/paperweight-userdev/src/main/kotlin/io/papermc/paperweight/userdev/PaperweightUserExtension.kt#L84 Would be nice to document this, basically you should set it when you are running gradle with a JVM too new to build a dev bundle
you can see the multi project test plugin branch for example uses
Can we ship the info in the dev bundle itself?
Tho I guess vendor selection would suck
at mods this guy is getting off topic

@pliant canopy 
I can do on Sunday
Wait wth since when is this in kotlin
Since the age of dawn, when birds started singing their first melody, and when foxes still said goodnight to rabbits; when humanity was a legend yet to be told by tongue and the dream of a powerful being had just begun...
Nah haha
I think all paperweight related tooling is Kotlin
sculptor, mache and paperweight at least
π π
@hard quartz whats "Took 7 minutes" 
a IDEA plugin i forget to turn off...
that's funny, I thought it might be an AI agent or something
So for the Astro version on the download pages, are we cool with the slight second delay in loading the download pages while the client fetches with the skeleton loader or should I try to make ISR work via some Cloudflare Worker options so that it just caches the page on the edge for 10 minutes or so.
haveing it prerendered would be neat
Alrighty folks, I haven't implemented any of the Cloudflare Worker options yet but we do have a staging URL now! Feel free to play around and see how it goes. https://papermc-website-staging.papermc.workers.dev/
its normal the first time the downloads/x take a little of time?
I think that's what he meant with the CF worker options are not done yet
^
https://papermc-website-staging.papermc.workers.dev/software/velocity/ show another javadoc version and throw 404
Looking.
and for waterfall the end of life banner its touching the logo (not sure how define this xd)
Is it only Velocity that releases Javadocs as a version without a snapshot tag in the URL?
the hangar image has different fitting and the navbar has different padding
other than that the landing page is exactly the same
Like Folio is broken too. The latest version is 1.21.8 but the latest javadocs for it is just 1.21?
maybe @small harbor can update the folia jd config
I can fix velocity, I can just cut the snapshot tag off of it cause I assume it'll never be in the URL.
Probably not the best way but it's a bridge we can cross when we get there.
I think the hangar image different fitting is just the responsive layout changes between how Next and Astro handle them.
Better?
Both of those fixes should be deployed. Looking at the navbar padding.
Now im eating and had the bad habit to use phone too i found this.
maybe but the way it is now has an empty space under the image, it doesn't fill the whole div
I'm looking into the downloads page ssr & caching
my ide is going crazy reindenting everything because there's no lint configs 
is it normal that team.astro takes so long to compile
oh
I'm trying to build so I can preview before pushing this
I think it works though
(just in run dev it won't have proper caching behavior afaik, hence why I want to use preview)
huh, preview errors even after build
β― bun run preview
$ astro preview
10:12:29 [@astrojs/cloudflare] Enabling sessions with Cloudflare KV with the "SESSION" KV binding.
10:12:29 [@astrojs/cloudflare] If you see the error "Invalid binding `SESSION`" in your build output, you need to add the binding to your wrangler config file.
10:12:29 [WARN] [adapter] Cloudflare does not support sharp at runtime. However, you can configure `imageService: "compile"` to optimize images with sharp on prerendered pages during build time.
[preview] The @astrojs/cloudflare adapter does not support the preview command.
Location:
/home/jason/Dev/PaperMC/website/node_modules/astro/dist/core/preview/index.js:44:11
Stack trace:
at preview (file:///home/jason/Dev/PaperMC/website/node_modules/astro/dist/core/preview/index.js:44:11)
at async runCommand (file:///home/jason/Dev/PaperMC/website/node_modules/astro/dist/cli/index.js:153:22)
error: script "preview" exited with code 1
sad
probably have to setup miniflare or smth
is there a reason there's a separate CF staging deployment instead of enabling PR previews in the wrangler config?
or I guess it needs to wait until merge to do that?
previews do not work with next.js
if I turn it on in wrangler and open a PR to this branch whats the chance it works
it is one of the whole reasons for switcing
it won't, no
should I just merge it to see if it works and it can be reverted if it's cooked
yea
main thing I'm not sure of is if I set the headers right, it's how cf docs say but idk if astro does anything funny or we have other settings
https://developers.cloudflare.com/workers/static-assets/headers/#custom-headers ok it looks wrong actually
I was on the pages docs instead of workers earlier
Hey @random pollen I'm back from lunch now. So far for the edge-cache I think you're on the right track but I think inside of Astro itself we can also make some changes. I think We should be able to also do a very lightweight middleware that interacts with the Cloudflare edge cache for the download pages.
what's the advantage of that vs just using the cache headers?
also I swear it works locally but the page 404s on the preview lmao
might be something with the wrangler config?
Yeah, I had the wrangler setup for static, not on-demand yet.
Also when I send that initial message to you I had not seen the last commit you pushed to the PR with the response headers set on the page.
It's starting to look more like what I was going to approach it with.
also I think we can do better than this for the 404 page 
(thats from the prod site)
I kinda like the style of this

Want me to push a change to the wrangler that I think might fix it?
yeah sure
We get logs and metrics for free since we're enterprise, right?
also I think we should setup prettier & eslint (sadly biome is not fully compatible with svelte)
Oh lord my day is ruined
CC: @opal flare
This is for enabling observability
We could just skip it too, doesn't make a difference to me.
I think Mini was talking about migrating to Oxlint?
Sucks that Biome doesn't fully support Svelte π
yeah it only deals with the contents of <script> tags
just stay with eslint
a little confuse here... this https://github.com/PaperMC/website/issues/136 is till valid or can be close?
Okay I just pushed a change to the wrangler that should hopefully fix.
I would say it can be closed when the astro pr is merged
If it doesn't, I'll start to do some more research.
ok triage not allow add the relation issue-pr 
oh, I'll add it then ig
Okay downloads pages are back
@random pollen You want to tackle eslint + prettier?
sure, I think we should also fetch the build list serverside
didn't realize it was separate
Yeah for sure if you want to add that into the download pages and update the svelte components for it, that'd be great π
I'll do linting first
But yeah the team page takes a little bit because it generates responsive image sizes for all the images on that page and serves them statically (a huge benefit over NextJS that does it at run-time)
80 / 2? Idk
I think 80 is a bit too short yeah
imo
I usually go with 140 or 160 
140
Obfusucate it 
140 then
ok 140/2spaces it is
Can you add in VSCode settings for run on save please?
I don't use vscode, you should add it
(using zed, would use IJ if svelte support wasn't so cooked)
somehow I didn't commit the correct scripts it seems
why did the worker build fail π
github pr explode.... great xd
@opal flare @small harbor could one of you please check the error log on cf
for some reason the linting changes broke the build
oh nevermind I found it
[@tailwindcss/vite:generate:build] `@source` paths must be quoted.
file: /home/jason/Dev/PaperMC/website/src/styles/global.css
Stack trace:
at file:///home/jason/Dev/PaperMC/website/node_modules/tailwindcss/dist/chunk-MC3TVECV.mjs:31:4363
at j (file:///home/jason/Dev/PaperMC/website/node_modules/tailwindcss/dist/chunk-MC3TVECV.mjs:3:1574)
at async vn (file:///home/jason/Dev/PaperMC/website/node_modules/tailwindcss/dist/chunk-MC3TVECV.mjs:37:637)
at async Ka (file:///home/jason/Dev/PaperMC/website/node_modules/@tailwindcss/node/dist/index.mjs:10:3433)
at async Object.transform (file:///home/jason/Dev/PaperMC/website/node_modules/@tailwindcss/vite/dist/index.mjs:1:2338)
error: script "build" exited with code 1
what did it do
is it just the line breaks π€
lol it is
That was a syntax difference of conversaion with React / Astro and React /Svelte
I can fix it when jmp is done with the linting stuff.
well this not hurts.. i think
am done with that, just pushed SSR for builds list
might add type checking though
(done with that now as well)
I guess a middleware could improve things a bit by sending loading skeletons again if there's no cached page and simultaneously caching the hydrated page serverside
to avoid the rendering delaying initial page load when there's no cache
but not sure exactly how to do that
It seems pretty fast now, good work.
yeah the HTML is all pre hydrated with builds and version info, but the initial latency is not great for me, varies from like 300ms to over 2 seconds
I don't see cf-cache-status in the response headers either
I do see the ones I added though
oh wait something is adding cache-control: max-age=0 to the request headers
I don't have disable cache ticked π€
oh it must be from refreshing duh
not sure if that behavior is ideal though
I guess it's good?
instructions for llm agents
I just symlinked it to the readme
(that way it won't try running npm commands)
CDN-Cache-Control is ignored by Cloudflare when used in _headers
Ahh
not using that anymore, we set them in astro
i.e. ```ts
export const prerender = false;
import Layout from "@/layouts/Layout.astro";
import SoftwareDownloadPage from "@/components/data/SoftwareDownloadPage.svelte";
import { fetchBuildsOrError, getProjectDescriptorOrError } from "@/utils/download";
Astro.response.headers.set("Cache-Control", "public, max-age=300");
Astro.response.headers.set("CDN-Cache-Control", "public, max-age=600");
const projectResult = await getProjectDescriptorOrError("folia");
const buildsResult = await fetchBuildsOrError(projectResult, false);
<Layout
title="Folia"
description="Download the latest builds of Folia, the experimental Paper fork with regionised multithreading."
keywords={["papermc", "minecraft", "performance", "paper", "downloads", "jar"]}
canonical="/downloads/folia"
<SoftwareDownloadPage
id="folia"
description="Download Folia, our new fork of Paper that adds regionized multithreading to the server"
experimentalWarning="Download experimental builds of Folia, our new fork of Paper that adds regionized multithreading to the server. Proceed with caution!"
client:load="svelte"
project={projectResult}
builds={buildsResult}
/>
</Layout>
ah you removed that header?
no, switched from using _headers to setting them on the page.astro
I'm going to test something real quick.
So going to test some Cloudflare edge fetching options and if that doesn't suffice, we might consider a lightweight middleware.
That's deployed. See if it feels like it loads faster for you now.
yeah it's not taking more then 300ms now
when cache is disabled in browser
I think that's acceptable
Cool. Yeah I told the worker-env when fetching to set an edge ttl for the worker fetching process itself and to force cache it for the time we set.
should I check the astro docs or do you know what this means already
Their devtools are kinda neat
Sec I'm pushing vscode settings
Yeah I know what it means. I can push fixes.
We might just ignore it for now, or I need to probably look at the docs for proper SVG usage, cause when you switch it to their image component for this:

btw I didn't realize the terminal was interactive until looking at the code
pretty cool
π
I vote for now we just keep it in the regular img tag and focus on other things.
fine with me
Pushing a fix for this now.
Alright, what's all left to get this into production?
is it rebased onto main?
Yes
I spent like 30 minutes doing that the other day
The nav spacing is still slightly off, I can't get it to match perfectly.
It's like maybe 5 - 6 px different.
Normal merge? Why not squash?
This one and the other
I'm pushing a change for that image on the home page for hangar. I made it 1-1 of what it is now. I didn't actually realize the scroll bar was showing now because of that change π
π
What pages are those two things on? I can try to fix them
The download page, the part about build Explorer
-# also the team data need an update? 
Fixed both of those extra spaces
Pushing
It might need updating honestly. I thought I fixed all the files in the stupid rebase I did.
Let me check
Looks like it's the same as https://github.com/PaperMC/website/blob/main/assets/data/team.json right now?
doc is new triage
-# yep, and not sure if another changes in the team?
also what if we just made new builds trigger a site rebuild
maybe using fills webhook system
that would solve slow first load when a region has no edge cache
@opal flare ?
maybe
Well it's going to be 1 person once in awhile that runs into that
Everyone else will see it fine
it wonβt happen as often once itβs in prod yeah
just seems more straightforward to βinvalidate the cacheβ when needed instead of time based
Fixed nav minus 1 px
I'm not fighting this 1px. It's much much closer to 1:1 now.
Will be deployed soon
I see a way to speed up the loads by like 30% too 
there's a duplicated builds request
Oh?
or maybe not actually
it might be duplicate but the gain looks smaller on second glance
There is duplicated but the cache is already warmed by the time the 2nd request comes through
So it's going to be minimal.
Could try to promise.all them to run them parallel?
possibly
can we pass a limit param to the fill api
not yet but we can pass a channel filter...
Can you just build a util function that wraps and returns the result and builds in a single function and them parallel fetch the backend data for that?
Cause in the get project descriptor it already fetches the builds so just return the builds from that initial call?
Something like
const { descriptor, builds, latest } = await getDescriptorAndBuilds("paper");
don't worry about limits and stuff, working on some changes to fill which will help with that
we will change it after that is done
Apparently we can also use a KV binding to warm all pops at once but I'm not 100% sure on that yet.
We're pretty dang close though I feel.
well once we have limit param we won't want to reuse it
since for the descriptor query we will limit 1
with channel filter
Oh sick so the extra speed will just come from the fill changes
then for build list we limit 10 and no channel filter
yeah mostly
I pushed what we can do for now
Awesome, ty.
Then I think... we're done minus a Team update?
Folia JD link is broken but I think we're waiting on Michael to push new ones?
Cause they were never deployed for 1.21.8, so, the site logic is right there for grabbing the latest version.
Oh I never did test on light mode. Did things seem better?
same issue when clicking the big download button's dropdown
Oh yes I never did address that, let me do that rq.
also @opal flare a multi-channel filter would be good so we could query for stable or recommended
Dear lord everything is so bright
can you open an issue for this, and an issue for limits? please
nice
Alright that's pushed. Let me go suffer and look through the rest of the site in light mode.
there are some sites I prefer in light mode these days - github and papermc.io are two of them
what about fill-ui.papermc.io
We need a cute paper chan bg on fill 
Found one last thing that needs fixed. Working on it.
Okay pushed a fix for that thing
I think it's pretty much ready to go now.
The sponsors missing image is no longer just a white circle and it matches with prod 1:1 for the initial as the fallback.
is it intentional that styling is different for 1 initial vs 2 initials
oh
not sure if this was a thing on prod but this hangar screenshot also looks a bit off
Checking
(looks like it needs the same fix as the landing page hangar screenshot)
Good catch. Fixed. Any others? I'll group em all in a single commit.
same thing on folia/waterfall/velocity
Hangar always causing problems 
Waterfall doesn't have a hangar one?
But I did fix for folia + velocity.
Pushed that
oops
can't you just wait until the dropdown closes before starting the transition
I guess I don't really get what transition you're trying to add
i.e. await tick(); requestAnimationFrame(() -> ... ; or smth
Am just messing around at this point
Not going to worry about that right now. I think we're ready for final checks and maybe riley pushes the button? π
eh I personally design the inconsistencies
UI changes quite a lot depending on the page
The only thing i find in the first navegation.. that little delay in the color change...
you mean when toggling experimental?
Ahh yeah I see that too
I'll fix it
That, but only the first time
fixed
Good catch π
well, fixed for the build list, I think the part with the big button was already there before my changes
it's much less noticable now though
I think it's fine for now. I don't see someone chewing us out for something that small.
Yeah only happen one time then cannot reproduce later 
the font on the paper page's description is a bit smaller than the other/the main site
Good catch
Lemme check
I see why
Fixed & pushed
@brazen moon Should be deployed. Can you confirm it looks correct now? π
yep perfect
Beautiful.
@random pollen @opal flare Anything else we need to get in before we go ahead and move this to production? Once that's done we can start working on the blog π
I think it's good for now. Riley or someone else with access had to change the Cloudflare configs.
Do we need to turn observability on in the wrangler config for our analytics or is that separate?
@small harbor
wrangler config
Kk I'll turn it on
@viscid thistle I snuck in a view transition https://github.com/PaperMC/website/pull/166/commits/b80e10dd2fd7b5a77caa23d0116f542be762da0a
Ooo
I have to enable throttling to fast 4g at least to see it 
Okay observ should be turned on via wrangler config now
When merge the name need to be change here not?
wrangler.jsonc
Ooo good catch. I'll have to coordinate that with riley.
That not affects later for future PR previews?
Unsure yet.
Running Astro on Workers is still kinda new for me. I still use Pages.
from my experience so far it's basically the same
idk what you mean by this
I mean if with all this then all the future PR can has the preview for them
once it's merged we should be able to get previews working yes
well for PRs from the PaperMC repo
not sure if the third party pages action for outside PR previews works with workers yet
Going to go make some dinner. Unless something else comes up this should be good to go. Just ensure you update the name in the wrangler config.
this one its a little hard and strange thing xd
a few avatars show a white line (second image) when pass the mouse over others avatars... if you pass the mouse over the avatar with the white line then fix for future...
How long you dig to find that one? π
just a coincidence when scroll xd
looks fine here π
Awesome π
oh no...
better close the staging page xd
the downloads per project not has the commit in footer
now close the page for stop find things xd
I'll take a look in the morning
microbump so I have an easier time finding this
Okay what am I looking for again?
I'm kinda confused
Oh nvm I see it
Did it used to? I wonder if it's because it's not pre-rendered and it's trying to pull env var from Cloudflare project at runtime.
Not entirely sure the proper fix for that right now. Need my coffee.
i think
I meant like on staging before we turned off pre-rendered
I think I have a fix actually, just need to bake it into vite's build step to expose during runtime.
ahhh dont remember if a see that part before xd
Fixed. Should be deploying shortly.
Looks fixed 
now is time for not touch/see and squash-merge xd
Just gotta wait on Riley π
Cause I think we have to coordinate the wrangler config name update.
i hope this comes in squash because that PR are many commits xd
what is left?
anything that ain't right
I think we just need to coordinate the wrangler config name change. Need to turn off the wrangler deploys so I can update the config name and then we move it over to dev and then I think @random pollen has some insight on how to ensure previews are setup properly once we get this merged.
We've been able to fix anything that gets brought up, so, I think it's pretty much ready to go? Any small tweaks that come up and be addressed as needed.
on fill-ui all I did was enable it in the wrangler config like I already did
Ahh okay cool
So yeah I think we just need to halt the previews real quick, update the wrangler name conf, merge the PR and ensure it deploys.
Which if you're ready riley, I can go ahead and get the commit pushed to update the wrangler name and then we can merge this in and deploy to production.
update the config, push, and then ping me and I will merge
Pushing now
I'm also going to drop the current blog branch once I save some things locally and I'm going to start a new one based on Astro.

Whatever it was for the previews
bunx opennextjs-cloudflare build
bunx opennextjs-cloudflare deploy
now that is a commit
website may break briefly
production is on https://github.com/PaperMC/website/commit/39b8f6a @viscid thistle
It's alive! Great work everyone!
now PR enabling previews since you forgot that part
its already enabled not?
no
jmp added "preview_urls": true, to the wrangler
oh he did?
Yes
I did not see that
then now the next things for website its the blog feature and update the teams.json not?
Yeah the teams.json should be a pretty easy update if anyone wants it.
I'll start working on a new blog branch here soon.
Now I get to write something up for the Astro showcase 
i mean i can just feels strange make that for self-insert xd
great job on the website
Really happy that we finally made this a reality. Was fun.
idk if it was like that before but the alignment bugs me, needs an items-center on the row
Ofc you wait to say that until after merge smh my head.
xd
could also probably use the assets CDN for the logos and stuff
I didn't have time before the merge!
ok
the renovate explode (?
https://github.com/PaperMC/website/pull/135#issuecomment-3418583763
expected
thats set in the bunfig.toml for security reasons
maybe there's a way to tell renovate to delay it's PRs as well
@hard quartz didn't see your PR first but I don't think it applies to all deps (specifically major versions)
true then maybe your PR match better...
i forget how renovate works for the rules... if takes all or what in this case
basically there are filters for what each rule applies to, the one you edited is grouping minor and patch updates
I made a new one that applies to all
will merge doc's team and jmp's renovate
whats this live content collections thing
sounds like maybe useful for the downloads page
seems like it
tbh at first glance I don't really get the difference from just normal clientside API calls or SSR
Didn't really think about the live content collections but it didn't fully make sense to put the two together. I know we'll use regular content collections for the blog.
I do need some insight from Riley on what all he plans to have in the blog posts. Do we want hero images? Here's some examples: https://docs.astro.build/en/guides/content-collections/#defining-the-collection-schema
optional hero image possible?
or actually
what we could do is use minecraft screenshots as the hero? something featuring a new addition?
/cc @last bear ?
I don't think we need proper hero sections, unless you just mean a small banner-ish thing with the title and a background at the top
-# I'll do some bikeshedding and throw in that it's not going to be a blog but news pages
yeah, same as like a minecraft update article with a hero at the top
iirc
is what glare had for next.js
Riley can you look on the Xenforo Admin panel side of things and see if you can do a bulk export of the Announcements section in either XML or JSON and I can try to put something together to convert them all to MDX to save us a bit of time?
Nah I don't think we should throw in something as large as those, we really rarely have news covering the actually visible content and otherwise they'd fit better into the actual body of the posts since they wouldn't be representative of the entire thing ever either
Sounds good.
no export options, apparently
Guh. I don't want to have to do this via SQL queries.
i'll whip up an export
https://forums.papermc.io/threads/1-21-7.1635/index.rss looks like RSS isn't going to help 
You might be able to generate me an API key and I can try hitting an endpoint.
For scopes I think it's node:read, thread:read, and post:read?
dm'd something
Well this is fun. Got things started with the blog and I went to go build the site and I'm getting Bun panics.
does it work if you comment the bun run part of bunfig
Checking.
Nope, but a different error this time at least.
Somehow getting a JS heap out of memory.
Okay well I fixed it but I don't know why what it was doing broke to begin with. I had a MDX file outside of a folder on accident.
Or not, it's crashing again. Anyways, I'm going to push what I have so far for the blog. I did not port over any styling from the NextJS branch but I did add in tailwind typography and I took the data that Riley got me and got it started on the cleanup.
If anyone wants to play with it: https://github.com/PaperMC/website/tree/feat/blog

link?
how firefox mess this...
first is chrome and second firefox in the latestVersionFrom
some form of sorting logic or?
chrome sort: 1.7.10,1.8.8,1.9.4,1.10.2,1.11.2,1.12,1.12.1,1.12.2,1.13,1.13.1,1.13.2,1.14,1.14.1,1.14.2,1.14.3,1.14.4,1.15,1.15.1,1.15.2,1.16.1,1.16.2,1.16.3,1.16.4,1.16.5,1.17,1.17.1,1.18,1.18.1,1.18.2,1.19,1.19.1,1.19.2,1.19.3,1.19.4,1.20,1.20.1,1.20.2,1.20.4,1.20.5,1.20.6,1.21,1.21.1,1.21.3,1.21.4,1.21.5,1.21.6,1.21.7,1.21.8,1.21.9,1.21.9-rc1,1.21.9-pre4,1.21.9-pre3,1.21.9-pre2,1.21.10,1.13-pre7
firefox sort: 1.7.10,1.8.8,1.9.4,1.10.2,1.11.2,1.12,1.12.1,1.13,1.13.1,1.13.2,1.14,1.14.1,1.14.2,1.14.3,1.14.4,1.15,1.15.1,1.15.2,1.16.1,1.16.2,1.16.3,1.16.4,1.16.5,1.17,1.17.1,1.18,1.18.1,1.18.2,1.19,1.19.1,1.19.2,1.19.3,1.19.4,1.20,1.20.1,1.20.2,1.20.4,1.20.5,1.20.6,1.21,1.21.1,1.21.3,1.21.4,1.21.5,1.21.6,1.21.7,1.21.8,1.21.9,1.21.9-rc1,1.21.9-pre4,1.21.9-pre3,1.21.9-pre2,1.21.10,1.13-pre7,1.12.2
good olllll sorting
Already opened a PR to fix it.
Not my favorite fix, but, it works.
Idk why it was only on Firefox and Safari
it looks weird, like how the project is being passed over
Wait don't merge
Why can't I convert this back to draft
I marked it as do not merge. I need to do some more testing.
It fixed Paper but it broke Velocity again.
Arenβt you guys using bun, why not just use their semver API?
It's a client-side fetch in the worker environment
I don't think that'll run under Bun
I don't like how we seem to be passing in what looks like a Project to a Record...
I would need to open my IDE but, 1am and on the wrong laptop π
hmm no preview working?
are you surprised about undefined behavior when your sort isn't transitive?
Okay I pushed something different. Hope that works better.
Idk anything about the previews not working, so, someone might have to pull locally to confirm.
yeah not sure if its the wrangler config or the worker...
i test in local for firefox/chrome/edge and looks its working
i feel safari work? i dont have that xd
Idk. I just installed Firefox to test this.
What is the sort even for?
To find the right version for javadocs
It's a semvar compare
Because we don't have consistency across projects
how do you mean?
Waterfall -> 1.21
Velocity -> 3.4.0-SNAPSHOT but the Javadocs is 3.4.0
Folio -> 1.21.8 is the latest but Javadocs never deployed for it
Paper -> 1.21.10
Yeah, but why is the actual sorting needed
Ah
I'm all for someone else to make improvements. I was just working with what I had π
Dear lord this was the stupidest freaking thing. Bun was crashing out because a few of the migrated posts had empty markdown link URLs. It just wouldn't tell me that on Windows. As soon as I switched to Linux, it told me the issue and it works again.
Blogs are pretty much ready to be worked on / styled by anyone. They are added in the sitemap and there's a rss.xml that is generated that will be under https://papermc.io/blog/rss.xml. All the blog pages just need styled and it's setup for MDX support so we can create some reusable blog components to throw into the posts.
now i remember.. not sure if @random pollen you know why the preview in website are not working? or is a thing for riley?
https://developers.cloudflare.com/workers/configuration/previews/ it's set in the config and I don't have panel access currently
Riley may have to go to the Worker project in the settings and enable the Preview URLs option. It looks like last month they changed it to opt-in.
We can go ahead and ping @opal flare and have him double check that whenever he's around.
@drowsy zinc btw I have updated https://github.com/PaperMC/docs/pull/642 to be up-to-date now and I think it is mergable. I also included the 1.21.9 system.properties changes (the descriptions are mostly copied from the mc wiki because I couldn't be bothered). Are you fine with me merging it?
I enabled it.
Hmm, okay.
So the next question, do we plan to use custom Astro components in our blog posts? And do we care if we have responsive-sized images in the blog posts? There seems to be a tradeoff between the two. With MDX (custom components), we have to do a little more work to have responsive images by default at build-time, if we drop the custom components, we can just just .md and I believe those are automatically responsive at build-time. I'm fine with either.
wait blog posts? Since when are those a thing? Thought they were posted iin the forum
or am I missing something and that's a new feature?
The forum is moving to read-only.
Oh, well that makes sense
so you are moving blogs from the forum to it's own thing?
I do not know the timeline for this ^^
that sounds nice
They'll be integrated directly into the website.
We haven't done much work on the styling yet, but, here's kinda what we have right now:
That should be possible.
I'd say the more we use it the more we'll end up customizing it, so probably good to have MDX support
Okie dokie
Okay cool so the same code works in Linux for this but running on Windows causes Bun to panic.
panic?
which is?
did you address my concern strokkur about 0 not meaning global for the config?
I was fooling around with the starlight-blog plugin in the docs codebase for 20 minutes, here's the patch if anyone needs a little inspiration (styling/layout/etc.) from an existing blog thing: https://pastes.dev/S8hbgg5dE2
It's related to using the Image component in the MDX. It works, just not on Windows env for some reason, it'll still build and deploy on Linux, just dumb. I will probably just go without the responsive images for now.
I hope? If you put it into the review, probably
If not, a quick patch never hurt anyone
(also, lulu, feel free to ping me whenever)
Is this still wrong? https://docs.papermc.io/paper/reference/spigot-configuration/#world_settings_default_end_portal_sound_radius
The number of blocks that the end portal opening sound is audible. Set to 0 for global (including cross-dimension). This setting has no effect if the gamerule globalSoundEvents is set to true, which is the default.
Yes it was in the review but was resolved before
