<sajcho>
SiFuh: (Wonder if a static build of a system is actually smaller than a dynamic.) I personally build a system without static libraries (except for exceptions) - size difference? I'll try it sometime.
sajcho has quit [Quit: Client closed]
lavaball has quit [Remote host closed the connection]
<remiliascarlet>
Static builds always result in way larger binaries, because everything it needs is baked into the binary itself.
<remiliascarlet>
But it does mean that regardless of where you run it, as long as the OS kernel and CPU architecture is the same, you can run it.
tilman has quit [Ping timeout: 252 seconds]
tilman has joined #crux
SiFuh has quit [Remote host closed the connection]
SiFuh has joined #crux
<ukky>
sajcho: thanks. My patch for non-static build is exactly the same. Nice additions to pkgutils in regard to multi-arch support.
sajcho has joined #crux
<sajcho>
ukky: Thank you. I added support for zstd and the binaries are built with the architecture flag in the package name.
sajcho has quit [Client Quit]
<ukky>
sajcho: exactly, adding arch in the pkg name is nice feature
<joe9>
groovy3shoes: thanks.
joe9 has quit [Quit: leaving]
brian|lfs has quit [Quit: Leaving]
ppetrov^ has joined #crux
sajcho has joined #crux
<sajcho>
Finally a success. Building firefox 121.0b5 on riscv64 using python 3.12 is done. 5 and a half hours of compilation. I'll have a beer and go to bed.
z3bra has quit [Read error: Connection reset by peer]
z3bra has joined #crux
<farkuhar>
ukky: grep for "getuid()" in the prt-get source code and you get no hits. It looks like jw preferred to delegate the uid tests to lower-level tools. I can see why stenur might have wanted the same delegation of responsibility to be applied in /usr/bin/ports (FS#1852), i.e., let the drivers or the filesystem perms handle it.
tarxvfz has joined #crux
tarxvfz has quit [Quit: tarxvfz]
<farkuhar>
one thing I find curious about jw's vision for future development is the claim that "code for aliasing" is in the wrong place (https://crux.nu/Wiki/PkgutilsIdeas#ntoc9 ). If aliases are only meant to indicate alternative ports that can satisfy listed dependencies without being an exact match, and no other command in the CRUX packaging ecosystem takes dependencies into account, then why is prt-get the wrong place for aliasing code?
<ppetrov^>
farkuhar, maybe aliasing should be on the "ports" level
<ppetrov^>
?
joe9 has joined #crux
zorz has joined #crux
<zorz>
hello ppl
<ukky>
farkuhar: wrt "getuid()", I think current implementation is correct, most CRUX users do not need extended functionality. And it was a good call to close FS#1852.
Guest81 has joined #crux
Guest81 has quit [Client Quit]
Guest81 has joined #crux
Guest81 has quit [Client Quit]
groovy3shoes has quit [Remote host closed the connection]
groovy2shoes has joined #crux
<cruxbot>
[opt.git/3.7]: mupdf: update to 1.23.7
<cruxbot>
[opt.git/3.7]: firefox-bin: update to 120.0.1
dim44 has quit [Remote host closed the connection]
dim44 has joined #crux
Guest9825 has quit [Ping timeout: 252 seconds]
Guest9824 has joined #crux
<SiFuh>
jaeger: Do you remember how was channel logging in #CRUX was enabled here
<jaeger>
I do not know, sorry. Maybe they have instructions on the site?
<SiFuh>
jaeger: I emailed them from the site asked if I was op and never replied since
<SiFuh>
All good, you don't know and I don't know now. Makes use right and they are wrong :-P
<jaeger>
I don't see much to go on but there's a line at https://libera.irclog.whitequark.org/ that says "If you want for this logger to be present on a particular irc.libera.chat channel, drop a letter to whitequark." with whitequark being an email hyperlink
<SiFuh>
Yes, I did
<SiFuh>
Then asked me if I was channel op and I said I am jaeger are channel ops
<SiFuh>
Blah blah blah and no reply since
<SiFuh>
I guess your name is bad?
<SiFuh>
Hahahaha
<cruxbot>
[core.git/3.7]: perl: update to 5.36.3
<farkuhar>
ukky: closing FS#1852 was perhaps the right call, but it does make you wonder why jw didn't follow Per's example and put the uid test inside the higher-level program. This lack of symmetry suggests that jw envisioned a different partition of responsibilities than the one that applies to ports(8) and /etc/ports/drivers/*
<farkuhar>
At the time jw was writing prt-get, there was already a getuid() call inside pkg{add,rm}, so maybe he didn't think it necessary to insert one in prt-get itself. The flexibility afforded by customizable settings for {make,add,remove,runscript}command was deemed sufficient to handle most use-cases.
<farkuhar>
So if I follow Mark Rosenstand's suggestion from 15 years ago and write a new subroutine 'prt-get sync', as an alternative to 'ports -u', there are two conflicting precedents I could follow. Either let the error message be generated by the lower-level tool (e.g. /etc/ports/drivers/httpup complaining about permissions), or test uid early and exit if not running as root (as /usr/bin/ports currently does).
<ppetrov^>
farkuhar, about your prt-get fork... would you consider the following: to log automatically (in /var/log/sysup.log or sth) the summary after prt-get sysup, e.g. "-- Packages updated", where build failed, etc
<ppetrov^>
i usually keep track of what is updated, even after revdep, but would be nice if it's done automatically. Not sure hom much such a log file would grow, though
<farkuhar>
ppetrov^: a log file wouldn't grow at all unless logging was activated by the administrator. If your feature request does get implemented, it would follow the existing precedent, namely that pkgmk processes don't generate logs unless you put "writelog enabled" in the config.
<ppetrov^>
wait, pkgmk can write logs? didn't know that
<farkuhar>
no, only the pkgmk processes spawned by prt-get
<ppetrov^>
ah, right
<ppetrov^>
i have not tried your fork yet, just managed to get the system up to date, including my own stuff
<farkuhar>
but if you want to do pkgmk logging yourself, pipe the output to tee -a
<ppetrov^>
well, it would make more sense for me to log after sysup, because that's when a bunch of stuff gets updated and later on i may wish to check how the updates were progressing
<ppetrov^>
actually, why am i explaining this? it'äs pretty obvious...
<farkuhar>
ppetrov^ and I are thinking aloud, to compete with the noise over in #crux-social
<ppetrov^>
farkuhar, I am not getting into another debate about political, scientific or whatever topic anymore
<r0ni>
about 5 xfce modules have had updates today, I've updated the xfce4 desktop repo accordingly https://github.com/mac-a-r0ni/crux-xfce4 (this is my pretend update bot, beep boop)
zorz has left #crux [WeeChat 4.1.1]
zorz has joined #crux
<ppetrov^>
r0ni, cool. Are you having problems with Xfce's terminal, littering the home folder with .bash-history-*.tmp files?
<r0ni>
i recall you complaining about said issue before, i'm not sure its xfce terminal doing it, yet it happens to me on slackware as well
<r0ni>
on my current box anyway... something crux/slack have in common, maybe xfce-terminal, I do know it's not specific to crux at least
<ppetrov^>
r0ni, slackware current?
<r0ni>
yes
<ppetrov^>
i guess it is a vte problem
<r0ni>
ahh that makes sense if it is
<ppetrov^>
i just guess
<r0ni>
i'm not sure what triggers the behavior though... can't get it to happen
<ukky>
farkuhar: 'prt-get sync' should not check for 'getuid() == 0', and just rely on sudo/doas or root login. Similarly, 'make install' does not force user to have superuser access to the system. This makes tool simpler. But again, my voice has little credibility and no vote points.
lavaball has quit [Remote host closed the connection]
<farkuhar>
ukky: I agree, it aligns better with jw's original design if we avoid any getuid() calls in prt-get itself. That's how I implemented the sync command so far, in my Perl-based rewrite.