Re: [csswg-drafts] CSS WG drafts server (#12054)

Note that I'm not advocating for me maintaining the current server(s) in perpetuity, in fact, I'd rather like to be able to drop that burden. I just want the WG to have consensus about any infrastructure changes with full knowledge of what may be lost before making changes.

Also note that W3C is currently looking into providing server hardware, my primary cost here is time, which is almost entirely spent keeping the legacy systems online in the face of crawlers. Keeping basic services online (like email, the wiki, and hosting static files) is trivial and I don't mind continuing that.

>> Draft servers for CSSWG, FXTF, and Houdini

> This we can kill by just moving to GitHub hosting the drafts. We already do that, I just redirect any GH urls to the "proper" draft.csswg.org URLs, which are then hitting your server currently. Should be trivial to just update DNS to point that domain to GH instead.

Note that the current GH build is not handling issue lists or markdown files, tho this could be added.

You're also not addressing the loss of the historical versions. The WG needs to decide if this is valuable (all the current historical versions are stored as (gzipped) flat files, I can provide the archive). I presume the GH workflow can also be augmented to preserve previous copies if that's desired. The main thing you'd lose is the ability to lookup by dated URL.

>> The IRC logbot (log.csswg.org)

> Now that we log our meetings to GH comments with github-bot, and do most of our discussions in issues (where in years past they'd often take place in the IRC), this is probably no longer necessary. If we want to keep it for historical purposes (which I think is valuable), moving a static version to a GH repo is probably sufficient.

Note that the logbot has logs for seven IRC channels, and is currently logging five. As far as I'm aware, it's still used by people collecting minutes. Github-bot doesn't collect the full transcript of a meeting as far as I'm aware.

Also, if this moves to a static version, someone needs to take the DB and generate those pages. It's not without labor cost.

>> The CSSWG Wiki (a Dokuwiki instance)

> I think we can move this to a GH wiki; this would simplify people's account management, and it's a low-traffic wiki anyway, mostly used to host meeting coordination info. I can do the migration if you send me the data files.

There's a lot more there than meeting info. IIRC there were some strong opinions over the choice of wiki software. Past discussions about moving to a GH wiki were not accepted.

>> An email server hosting an archive of www-style available via IMAP.

> Unclear value here, since the W3C archives exist. (mailman archives suck, but at least they exist) We can probably kill this?

This was explicitly requested in a WG F2F years back. I don't know if anyone still relies on it, but it takes up none of my time to maintain as it's fully automated. It's just a mail account and a Sieve filter.

>> Shepherd, which includes the spec parsers that provide (provided?) anchor data to Bikeshed.

> Definitely "provided", I moved off of Shepherd for hosting those a few years back. Bikeshed's definitions are now handled by WebRef, run by the W3C.

Note that the anchor DB is also used by the test annotation script. It's unclear to me if this is still being used (pretty sure I saw it turn up on a spec recently), and if so, how relevant it is as I imagine the test data is somewhat out of date. If this (and the harness) goes away, the annotation script should silently fail in any specs where it's still present.



-- 
GitHub Notification of comment by plinss
Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/12054#issuecomment-2849474518 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Sunday, 4 May 2025 22:32:55 UTC