07:46 <ehawkvu> illiliti: emacs 29.1 for community - https://0x0.st/H_Hy.PGgLFC

08:13 <illiliti> what does esources file do?

08:13 <illiliti> > # Determine whether we are the -nox package or not...

08:14 <illiliti> i wouldn't do it

08:15 <illiliti> also isn't ./configure script smart enough to make all these pkg-config checks without assistance?

08:16 <ehawkvu> the esources file is my own thing - i can rm that

08:17 <ehawkvu> and the configure script does *not* check for libgccjit automatically so that option at least has to stay

08:17 <ehawkvu> and fair enough on the -nox check, i figured if i could share the build script i should

08:19 <ehawkvu> more context on esources: https://github.com/ehawkvu/kiss-personal/blob/master/bin/kiss-echecksum

08:19 <ehawkvu> i've been using it to much success over in kiss-xorg

08:23 <illiliti> i'm kinda against using kiss in build scripts, so if there is a more agnostic way to check for it please remove kiss in favor of it

08:25 <ehawkvu> hmm

08:25 <illiliti> esources file - sorry but non standard extensions are definitely no go

08:30 <ehawkvu> illiliti: figured that

08:31 <illiliti> if there is no better way to check for libgccjit don't bother. i won't block your patch because of that

08:32 <ehawkvu> I have a hacky solution - https://0x0.st/H_H0.mMbaJM

08:33 <ehawkvu> When you compile and run that it'll tell you whether or not libgccjit is installed

08:33 <ehawkvu> but I really don't like it lol

08:36 <illiliti> me too

08:36 <illiliti> i feel it could be simplified but not sure how

08:37 <ehawkvu> Here's the manifest of libgccjit - https://0x0.st/H_H5.BBAJaj

08:40 <ehawkvu> also should -nox be committed? I don't mind adding it but i think most ppl will be using the gui version anyways

08:42 <illiliti> yeah we shouldn't rely on the fact that it's exists in /usr/lib

08:43 <illiliti> personally i think that -nox is redundant because we have kiss-fork which can be used to tweak package how you like

08:43 <ehawkvu> right

08:45 <illiliti> case $("$CC" -print-file-name=libgccjit.so) in */*)

08:48 <illiliti> basically if $CC produces output with path, then library exists. if no, then doesn't exist

08:48 <illiliti> we use this trick in main repo, so there is nothing wrong to use it in community as well

08:48 <ehawkvu> sweet

08:48 <ehawkvu> should I just check for the exit code?

08:49 <illiliti> no, you should check output. exit code should be ignored

08:49 <ehawkvu> ok

08:51 <illiliti> https://codeberg.org/kiss-community/repo/src/branch/master/extra/rust/build#L118

08:54 <ehawkvu> new patch - https://0x0.st/H_Xr.FCecjj

08:54 <ehawkvu> also ty illiliti

09:09 <illiliti> merged, thanks!

09:10 <illiliti> btw why emacs is so prone to segfaults as you noted in build script?

09:13 <ehawkvu> that is something I have been trying to find out myself tbh

09:14 <ehawkvu> based off of what ive read in the repo & my own testing, all of the segfaults

09:14 <ehawkvu> happen with the executable `temacs'

09:14 <ehawkvu> which, from what I understand, is the pure C version that is used to bootstrap the full elisp version

09:15 <ehawkvu> but for some reason, that C version is very picky with how it likes to be built

09:15 <ehawkvu> i remember reading the ASLR stuff in etc/PROBLEMS

09:17 <ehawkvu> lol: https://github.com/emacs-mirror/emacs/blob/1e8322bb26e4945de460780168732250bbd083d0/configure.ac#L6524

09:17 <ehawkvu> apparently they don't know either

09:17 <ehawkvu> lmao

09:20 <illiliti> amazing shit

09:21 <ehawkvu> what's even better is that you can build with -O3 -march=native if you use clang

09:21 <ehawkvu> I haven't tested trying to use -flto though

09:21 <ehawkvu> that may be the breaking point

09:22 <ehawkvu> and iirc it doesn't require the -no-pie flag

21:43 <zlib> hi my favorite people i have a really niche question about kexec and heads if anyone knows about it? i tried asking in coreboot but nobody knew at all

22:52 <zlib> never mind the lovely reddit man helped x