02:10 <jaeger> I wish I had saved that qtwebengine package last time I built it... going to this time, heh

02:39 <SiFuh> I have saved mine, but don't use it anymore.

04:01 <jaeger> qtwebengine is the first thing that made me want more than 8GB RAM in my laptop

04:01 <jaeger> guess I'll build it on some other host

10:15 <beerman> thanks jaeger , i indeed don't look there..

15:44 <stenur> i have switched to overcommitting tmpfs that backs my PKGMK_{TMP,WORK}_DIR.

15:45 <stenur> before i had used a /var/tmp/ports/.. thing, which was ok but shitty, sometimes stale directories were laying around, i never made my way to dig into pkgmk

15:46 <stenur> then i reread tmpfs Linux documentation, and switched to overcommitting. i thought that does not work, do not know why

15:47 <stenur> so now everything is there and i just do not care no more. my NVME is so fast (r+w ~1.5 GB/sec) that .. actually i do not even know how fast RAM is nowadays.

15:48 <stenur> "when i was hot" i think it was about ~333 MB/sec on an Athlon?

15:48 <stenur> na..too slow, hmm. or not??

15:53 <stenur> well maybe 500 MB/sec says Wikipedia. if it's true.

17:24 <jaeger> I ended up uploading a built qtwebengine package to https://crux.ninja/tmp/ in case anyone else runs into the annoying cores vs. memory issue

18:40 <farkuhar> Like SiFuh, I gave up on qtwebengine a while ago. It looks like the dependent port qutebrowser is slowly transitioning to qt6-webengine, but that won't necessarily improve the memory usage during builds.

18:42 <farkuhar> https://github.com/qutebrowser/qutebrowser/tree/qt6-v2 if anyone wants to try a branch that supposedly builds without qtwebengine.

20:24 <stenur> I really want to beg people of core to switch libarchive back to autotools. With cmake threading is not supported.

20:24 <stenur> i have opened issues, now on github, too

20:44 <darfo_> qt6-webengine is almost as bad and will likely be as bad as qtwebengine after a time

20:45 <darfo_> i've got 16GB RAM but qtwebengine thrashes swap if I let it use to many threads.

20:45 <darfo_> 2 threads takes a long time to compile but the system is still usuable during

20:46 <darfo_> but meson is arrogant, it doesn't respect pkgmk.conf JOBS or MAKEFLAGS

20:46 <darfo_> so I put a wrapper around meson and force it to use -j 2

20:57 <jaeger> yeah, I ended up just disabling 16 of the cores with chcpu

20:58 <jaeger> meson/ninja didn't want to cooperate with me

21:02 <darfo_> didn't know about chcpu, thanks!

21:27 <jaeger> np

21:57 <pitillo> hey, can someone check on x86 if last libarchive update breaks pkgutils build?

23:07 <farkuhar> pitillo: pkgutils still compiles successfully for me, with the latest libarchive (3.6.2-1).

23:56 <jaeger> pitillo: seems fine for me as well