[08:39] <tomwardill> wgrant: do you know much about our CSS build pipeline and why it's structured the way it is?
[08:39] <tomwardill> something something 'components are odd'
[08:40] <wgrant> tomwardill: I don't. The most recent serious contact I had with our CSS was before I was with the company, laughing at the silly idea of redesigning the whole thing in one month.
[08:40] <tomwardill> ah, right
[08:40] <wgrant> ... they did it and you see the result.
[08:41] <wgrant> Colin might have touched it a bit more recently.
[08:41] <wgrant> But I haven't made non-trivial changes in a decade.
[08:41] <tomwardill> we appear to rewrite all the URLs in the CSS, which means the plan to just do a straight node-sass concatentation doesn't work.
[08:41] <tomwardill> I was wondering if something along the lines of a collectstatic approach might be better
[08:42] <wgrant> Is the rewriting thing just for absolute icing paths, or something else?
[08:43] <tomwardill> the former I think
[08:43] <tomwardill> at the point of concatenation, we grab all the CSS files from components and yui, then rewrite any urls in the CSS (for icons, etc) to point at the right place
[08:43] <tomwardill> then do the concat
[08:44] <tomwardill> but there's also something to do with revision numbers that I don't quite understand yet
[08:46] <SpecialK|Canon> pure guess but prepending revision number to output paths so users get stable UI experience?
[08:47] <wgrant> Right, outside dev we always prefix media with a path containing revinfo and Expires: 5eva
[08:47] <wgrant> (there should be an RFC for that)
[08:48] <tomwardill> yeah, that makes sense
[08:48] <tomwardill> doesn't always seem to apply though
[08:48] <tomwardill> I may well be reading the access logs wrong however
[08:51] <wgrant> Some HTML or JS likely uses non-revved paths because it's lazy
[14:04] <tomwardill> before swearing a lot and trying to work out why your fixed code is still generation 404s, ensure that the file actually exists on disk and that you haven't rm'd it and not regerated it
[14:04] <tomwardill> I'M LOOKING AT YOU, bg_steps-estatus.gif'
[14:07] <ilasc> :)
[14:07] <ilasc> nice one
[14:17] <tomwardill> can someone re-review https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/384706 given the last commit is a change in behaviour and is new since ilasc looked at it last.
[14:18]  * pappacena checking
[14:21] <tomwardill> pappacena: it might be worth starting the branch (via checkout, make clean, make run) and clicking some pages
[14:21] <tomwardill> and checking the logs for 404s
[14:21]  * pappacena 👍
[14:21] <tomwardill> entirely possible it's still broken and I've just missed some
[14:35] <pappacena> it depends on this one, right? https://code.launchpad.net/~twom/lp-source-dependencies/+git/lp-source-dependencies/+merge/384643
[14:36] <tomwardill> oh, ydx
[14:36] <tomwardill> *yes
[14:36]  * pappacena pulling it!
[14:42] <tomwardill> pappacena: you may need to apt upgrade nodejs too
[14:43] <pappacena> Should it work with `$ node --version` == v8.10.0 ?
[14:43] <pappacena> because apparently it worked oO
[14:53] <tomwardill> ah yes, that is the updated version
[14:53] <tomwardill> (origianl is v4)
[15:03] <tomwardill> SpecialK|Canon: I have no idea what longopts is
[15:03] <tomwardill> otherwise, maybe?
[15:03] <SpecialK|Canon> tomwardill: ah sorry! --help vs -h
[15:03] <tomwardill> ah, fair
[15:04] <SpecialK|Canon> (I had to look up `ln -n`)
[15:05] <tomwardill> oh, I wasn't going to change that one (it's copy+paste fromt he block above it)
[15:05] <SpecialK|Canon> fair enough!
[15:05] <tomwardill> there's perhaps a job in going through the Makefile for that though
[15:06] <SpecialK|Canon> Sure, no biggie, I just like/encourage the general convention
[15:06] <tomwardill> I've changed the one on the node-sass line to `--recursive` over `-r` though
[15:17] <tomwardill> well, this is a fun diff, but transcription to SCSS: https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/384834