<farkuhar>
Romster, "incompatible with i386 output" pointed me in the right direction, my pkgmk conf was not exporting the correct CFLAGS. I fixed the typo and got a successful build; no need to blame libbsd!
<Romster>
yeah not everything is some library's fault
<ahmadraniri[m]>
hello, any (full) wayland user here ? I'm using dwl wayland compositor and I can't make firefox to launch. I use MOZ_ENABLE_WAYLAND=1 option but still no luck.
<jaeger>
Not a wayland expert here but firefox works for me on it without specifying the env var... any errors when you run it?
<ahmadraniri[m]>
the important one is "can't open open display", it's logical because I use wayland compositor and not X, but it should launch when I set MOZ_ENABLE_WAYLAND= 1, like other distros do. I'll investigate further.
<jaeger>
May need to rebuild mesa/gtk3 for it if they weren't built with wayland support
<jaeger>
i.e. if they were installed before wayland
<jaeger>
though it's also possible I'm mixing this up with xwayland
chrcav has quit [Ping timeout: 248 seconds]
chrcav has joined #crux
mbarbar has quit [Ping timeout: 268 seconds]
mbarbar has joined #crux
SiFuh has quit [Ping timeout: 265 seconds]
ppetrov^ has joined #crux
SiFuh has joined #crux
ppetrov^ has quit [Quit: Leaving]
Guest5077 has joined #crux
Guest5077 has quit [Client Quit]
<farkuhar>
ahmadraniri[m]: the mesa/gtk3 packages on the installation media were not built with wayland support; you have to rebuild them as jaeger said. Your error message about a missing display suggests that you didn't install xorg-xwayland, which could have served as fallback in the event that mesa/gtk3 were not recompiled.
<ahmadraniri[m]>
<farkuhar> "ahmadraniri: the mesa/gtk3..." <- aight, thanks, will try it soon.
<farkuhar>
SiFuh: i might have created a duplicate page before the transition between namespaces (Sandbox -> Main) was complete.
<joshsmt>
Is it normal for Firefox not to compile?
<farkuhar>
Hello joshsmt! Are you building it with pkgmk directly in the ports tree, or with prt-get?
<joshsmt>
prt-get
<farkuhar>
And you're using the depinst command? ('prt-get install' does not resolve dependencies automatically)
<SiFuh>
farkuhar: Yeah, it came up in google though
<joshsmt>
yes
<joshsmt>
Is it because of the glibc update? I haven't updated it yet
<farkuhar>
joshsmt: can you paste somewhere your build log for firefox?
mkf has left #crux [so long and thanks for the fish]
<joshsmt>
where are the prt-get logs?
<jaeger>
It's never normal for a port to fail to compile, for what that's worth... not much use for intentional failure in port builds
<farkuhar>
look for the line that starts with "logfile" in /etc/prt-get.conf, that will tell you where they get saved (if writelog enabled)
<SiFuh>
Location will be ->> /var/log/pkgbuild/<pkgname>
<joshsmt>
yeah I didn't enable
<SiFuh>
farkuhar: I haven't looked at 3.7 yet is the logging still similar to 3.6?
<farkuhar>
on a fresh installation? weird, I thought "writelog enable" was one of the defaults that beerman changed for 3.7
<joshsmt>
yep
<SiFuh>
Thought that was runscript
<jaeger>
you can also just redirect your output to a file, like 'pkgmk 2>&1 | tee /tmp/build.log' or similar
<SiFuh>
... jaeger and his fancy commands :-P
<jaeger>
I personally don't have prt-get logging every build but now and then I want to save a log for an individual port
<SiFuh>
I have overwrite and rmlog_on_success set
<joshsmt>
i gonna compile again
<joshsmt>
anything I send the output here
<SiFuh>
NO!
<joshsmt>
what
<SiFuh>
use something like dpaste
<SiFuh>
Or what jaeger recommends
<SiFuh>
I can never remember the name of the URL
<joshsmt>
paste.rs?
<joshsmt>
termbin
<SiFuh>
Yeah that's fine
<SiFuh>
ix.io
darfo has quit [Ping timeout: 265 seconds]
<farkuhar>
`prt-get ls firefox | grep -c patch` shows 36 patches for our firefox port. I can't remember any other port having so many patches.
<SiFuh>
farkuhar: I was 'duckduckgo'ing CRUX and encryptiong and the CRUX37-Encrypted turned up at the top of the results. So I thought I'd question and mention it.
<joshsmt>
it is funny
<joshsmt>
because most of these patches are for musl
<farkuhar>
SiFuh: thanks. If pmwiki allows me to delete one of the dups, I can reduce by 1 the number of duckduckgo hits, just so your search results are cleaner next time.
ivandi has quit [Quit: WeeChat 3.6]
<SiFuh>
farkuhar: they are not the same, they are similar
<SiFuh>
CRUX37-Encrypted <-- this one seems to be the one that is your first
<SiFuh>
CRUX3-7-Encrypted <-- This one seems to eb the release
ivandi has joined #crux
<SiFuh>
But I can't be sure. So I really can't help :-P
joshsmt has quit [Quit: Client closed]
<farkuhar>
SiFuh: whichever page the Handbook links to, that's the one we should leave in place. The other one doesn't need to stick around.
<farkuhar>
Oh no, we lost joshsmt. I was looking forward to reading a firefox build log.
<farkuhar>
It's on the ISO but not in core. Funny you should ask, there was a Flyspray task that was left open for the longest time, even though cryptsetup has been on the ISO for several releases already.
ivandi has joined #crux
<SiFuh>
Probably from my whinging
<SiFuh>
farkuhar: So under Instructions, it might be wise to include cryptsetup in the list of required packaged
<SiFuh>
Extra packages: lz4 if you choose this compression mode for the kernel, dracut and lvm2 to access the logical volumes when booting
<farkuhar>
SiFuh and jaeger were blaming dav1d for the kodi build failure that brian|lfs experienced, but then brian|lfs offered to "nuke dav1d" from his system since he "never got firefox to compile". I wonder if joshsmt is seeing a similar dav1d error. You'd think that after 36 patches the firefox source tree would be in better shape, no?
<SiFuh>
Heh, no half way with you
<SiFuh>
Seems a lot more people are using torrent to get CRUX 3.7 than 3.6 or 3.6.1
<jaeger>
I was joking about the similarity between 'dav1d' and 'david', was focused on the dvdread/dvdnav issue, though
<SiFuh>
jaeger: And mine was a suggestion 'might' if I remember correctly was the word I used. Like i said "No halfway for farkuhar" :-P
<SiFuh>
Welcome to the world of assumptions verses presumptions
<farkuhar>
actually SiFuh wrote "Looks like dav1d is missing" (18:06) "Or exists and is screwing with his build" (18:08)
<SiFuh>
03:05:16 [SiFuh> brian|lfs: The existence of dav1d might be the cause
<SiFuh>
Hahaha
<farkuhar>
SiFuh's IRC logs are prepending *local* timestamps, but without providing the offset from UTC. Maybe he can help us debug the Gitea webhook, with all that experience translating timestamps in the absence of an explicit offset.
<SiFuh>
farkuhar: Do you mind to remove my poorly spelled name as well from the Encrypted page? Not a fan of credit
<farkuhar>
okay, will do.
<SiFuh>
The time stamp I saw was an empty value
<SiFuh>
The one ending in Z right?
<SiFuh>
farkuhar: prefer CRUX to carry my work and not me, for the most part.
<farkuhar>
nevermind, jaeger just clarified in #crux-devel that the timestamp issue has nothing to do with offset from UTC.
<SiFuh>
As I said, it looks like an empty value
<SiFuh>
I don't know anything about gitea or json
<SiFuh>
Noing web based stuff it probably needs some module to know what time it is, or it will just keep printing the same time and not tick over.
<farkuhar>
joshsmt: yes, it looks like Romster only pushed his commit to the 3.6 branch, not yet 3.7
<joshsmt>
yeah
<SiFuh>
stenur: Welcome to the beerman ignore list. Nice to have company now ;-)
<joshsmt>
should i wait or do i update this by myself?
<farkuhar>
joshsmt: it can't hurt to copy the wasi-libc++ Pkgfile from the 3.6 branch for a manual update. No need to wait for Romster if you're really curious about what's causing the firefox build to fail.
<joshsmt>
ok
stoffepojken has quit [Quit: ZNC 1.8.2 - https://znc.in]
stoffepojken has joined #crux
<farkuhar>
SiFuh: what did you do to get on beerman's ignore list?
<farkuhar>
joshsmt: sorry to hear that. Could you share an updated build log, or is it failing in the same place?
<joshsmt>
wasi-libc++ is failing in this case
<joshsmt>
it could be some wasilibc++ dependencies that need to be updated as well
<farkuhar>
i'm surprised that i got a successful build of wasi-libc++ myself, because according to prt-get one of its dependencies wasn't yet installed. Anyway, my build log: http://ix.io/4c2C
<joshsmt>
weird
<farkuhar>
actually a footprint mismatch with MISSING would qualify as an unsuccessful build; the absence of dependency wasi-compiler-rt might be responsible for the missing files.
<farkuhar>
so a proper test would involve creating *two* port dups: wasi-compiler-rt and wasi-libc++. Give both of them a version bump to 15.0.1 (copying from the CRUX 3.6 branch if needed), and see if that allows firefox to build for you.
<joshsmt>
o wasi-compiler-rt compiled successfully, now wasi-libc++ keeps failing
<joshsmt>
same thing
<farkuhar>
joshsmt: what else is in the deptree of wasi-libc++? Anything that hasn't been updated since your last run of ports -u?
<joshsmt>
clang, compiler-rt, llvm, cmake
<joshsmt>
the only one that updated was cmake
<joshsmt>
but i already updated it
<farkuhar>
if even one of those pieces of the toolchain is out of date, that might explain your build failure. Update all the llvm ports to 15.0.1 before proceeding to wasi-libc++ and firefox.
<joshsmt>
llvm it's already on 15.0.1
<joshsmt>
and llvm dependencies are up to date
<joshsmt>
I think it's better to wait for the commit in 3.7, otherwise I'll end up breaking something
joshsmt has quit [Quit: Client closed]
<cruxbot>
[core.git/3.7]: dhcpcd: fixed source()
<cruxbot>
[contrib.git/3.7]: prometheus-node-exporter: initial commit, version 1.4.0
<cruxbot>
[contrib.git/3.7]: prometheus-node_exporter: renamed to prometheus-node-exporter