06:59 <SiFuh> 06:53 <a1eko> Hi, crux ports page seems to link 'P F S' in Files to the old non-existent crux-3.6 for core, opt, xorg and contrib.

09:00 <beerman> jaeger: good call

10:39 <beerman> gettext can link automagically to opt/libunistring and man-db can link to libseccomp

10:39 <beerman> that needs to be fixed

11:10 <farkuhar> Re: gettext automagically linking to opt/libunistring, there's also the footprint mismatch that ahmadraniri reported in #crux (MISSING /usr/lib/libgettextlib.a), which i was able to reproduce on one 3.7 system but not another. Curiously, opt/libunistring was installed on both systems.

12:49 <beerman> jue: cheers for fixing up gettext and man-db

12:50 <beerman> i am on my round to update a few romster ports again

12:54 <jue> beerman: np, thanks for your work on romster's stuff

12:54 <beerman> somebody got to do it...

12:54 <jue> looks like he is _not_ back again

12:55 <beerman> eh... if I knew

12:55 <beerman> you can find him streaming on twitch if you want

12:55 <beerman> ¯\_(ツ)_/¯

14:15 <jue> beerman: problem with new gnupg: if nettle is installed and openldap is not the build breaks, works with 2.3.7 for me

14:16 <jue> beerman: adding --disable-dirmngr is a 'fix'

14:16 <beerman> heh, I will look into it

14:20 <beerman> https://gitweb.gentoo.org/repo/gentoo.git/tree/app-crypt/gnupg/files/gnupg-2.2.40-fix-no-ldap-build.patch maybe this solves it already

14:25 <beerman> jue: inside my container I fail to reproduce the error, however, it also builds fine with the patch installed and nettle installed while openldap is not. can you verify the patch solves it for you?

14:25 <beerman> err. shouldn't look into another terminal while typing. it builds just fine, with or without the patch is what I meant to express :D

15:24 <jue> yep, works with the patch

15:26 <jaeger> Did anything change with php on crux.nu very recently?

15:26 <beerman> not to my knowledge

15:27 <beerman> i updated gitea today

15:28 <beerman> jue: will bump that in a bit then

15:44 <jaeger> I think this portdb issue is caused by php 7.2 treating some warnings as fatal

15:44 <jaeger> PHP Warning: Declaration of RepoList::toHTML($repos) should be compatible with PortDb::toHTML($start, $headers, $items, $ports = -1) in /var/www/htdocs/portdb/index.php on line 330

15:45 <jaeger> There's a lot of funky OOP in this code and now it seems to be failing

15:47 <farkuhar> I can confirm that the patch from the gentoo gitweb fixes the build when "nettle is installed and openldap is not". The patch allows some (new) dirmngr files to appear in the footprint, which jue's "--disable-dirmngr" fix would have removed from the set of build targets.

15:59 <beerman> farkuhar: thanks

15:59 <beerman> jaeger: no idea honestly. I haven't been working on crux.nu for weeks now tbh

16:00 <jaeger> well, I'm no php expert, maybe jue has a better idea

16:01 <beerman> maybe we should try and rebuild the php version we are currently using against current libs? I did update a few things weeks ago

16:01 <beerman> carefully, but who knows, maybe that still broke something

16:03 <jaeger> based on the timing of a1eko's and SiFuh's comments this would have broken last night

16:04 <jaeger> To be fair I'm not sure if this is the cause... it says warning but the function which sets up those P F S links doesn't even get called, it appears

16:08 <beerman> https://crux.nu:82/portdb/?a=repo&q=core

16:09 <beerman> works with nginx

16:09 <beerman> it seems

16:09 <beerman> at least the links are there, still linking to 3.6 files

16:11 <jaeger> I updated the version in the one in htdocs, guessing nginx is using a different copy

16:11 <beerman> i literally just copied it over and made it available :D

16:12 <jaeger> with that said, the portdb php code is funky and could do with a rewrite, probably

16:12 <jaeger> I wonder if anyone even cares that it can produce XML

16:13 <beerman> i copied the latest portdb folder again and now the links are gone

16:13 <jaeger> odd

16:15 <beerman> personally i don't care for xml output

16:15 <beerman> is something recursing badly?

16:16 <beerman> i mean, it looks like a clean pattern but probably nothing you want to see in a log file :D

16:16 <jaeger> oops, that one's on me, sec

16:18 <jaeger> I left a var in that I meant to remove after testing

16:18 <jaeger> ... and now they work again. what the hell?

16:18 <beerman> working on nginx again too

16:18 <beerman> with the correct link as well

16:19 <jaeger> the error image you just posted was my fault but I have no idea why the original breakage occurred

16:19 <beerman> heh, funky indeed it seems :D

18:35 <stenur> that --disable-documentation in btrfs-progs is a really bad thing; couldn't you at least wrap it up like i do for contrib/ipcalc?

18:36 <stenur> for example mkfs.btrfs --help is simply not sufficient compared to Documentation/mkfs.btrfs.rst

18:40 <stenur> i absolutely concur to not add these tremendous requirements onto CRUX builds for documentation purposes, upstreams became antisocial regarding this, but here these are .rst format and can easily be read in plain, even if it is ugly. More ugly is not to have documentation and the need to unpack a ball to read it.

21:26 <farkuhar> beerman: on your round to update a few romster ports, the gcr update introduced a new dependency (gtk4) that wasn't mentioned in the commit msg.

21:28 <farkuhar> So I dutifully went ahead and built gtk4, but it wasn't on a clean container, so the CUPS print backend was automagically preferred over lpr, and I hit a footprint mismatch: http://ix.io/4dh6

21:29 <beerman> farkuhar: happens.

21:29 <beerman> yeah thats actually known

21:30 <beerman> i recently made my script use that maintainer-clean-footprint script that rust and others have so i should just use it to rm that file from the footprint