00:02 <cruxbridge> <tim> blackfrog78: could you create a ticket upstream at utill-linux about it?

00:12 <blackfrog78> I am not sure if util-linux with its pam_lastlog2 is the right place to fix it. /bin/login from the shadow port can write a wtmp entry under some circumstances (no pam) as ukky wrote. I just tried a more recent version of shadow (4.15.2) but it seemed to behave the same. Maybe I should report it to shadow upstream?

00:13 <cruxbridge> <tim> sure, maybe thats the better place

00:38 <blackfrog78> Issue created: https://github.com/shadow-maint/shadow/issues/1026

00:46 <cruxbridge> <tim> cheers, let's see then :)

10:24 <ppetrov^> some how a got a bunch of gcc files (e.g. usr/lib/gcc/x86_64-pc-linux-gnu/12.4.0/plugin/) already present _before_ gcc is updated

10:28 <ppetrov^> is it possible that they were not installed only to $PKG but directly (at package build time)?

10:41 <blackfrog78> ppetrov^: Did you need to use pkgadd -f to update the gcc package? I could update it without pkgadd -f.

10:44 <ppetrov^> i had to use the -f option, yes

10:44 <ppetrov^> otherwise it complained about a bunch of files already present

10:45 <blackfrog78> When I did the update to gcc 12.4.0 yesterday this was not the case for me.

10:54 <cruxbridge> <tim> for me neither on either host so far

11:26 <remiliascarlet> Why not use Clang instead? Works pretty much the same as GCC, and you'll be just as miserable compiling Clang as you would be compiling GCC.

11:27 <ppetrov^> remiliascarlet, because a lot of ports require gcc?

11:27 <remiliascarlet> Or consider using a compiler that doesn't suck, like scc or tcc, even if it means that 99% of the ports will be uncompilable, because these compilers demand your code is correct.

11:28 <remiliascarlet> ppetrov^: Symlink GCC to Clang to make pkgadd shut the fuck up about having to install GCC first.

11:32 <ppetrov^> just ln -s Clang to gcc?

11:33 <ppetrov^> actually I do not mind gcc itself, i was just wodering why the port behaves the way it does (here). That's all. nevermind, I used -f, so fuck it

11:33 <remiliascarlet> Yes. because if I remember correctly, CRUX simply looks at your /usr/bin directory, and doesn't otherwise care if it's actually installed or not.

11:34 <ppetrov^> gcc is partof core and I feel uncomfortable removing stuff from there

11:34 <ppetrov^> i also need gcc-fortran

11:34 <ppetrov^> a bunch of R stuff needs it

11:59 <blackfrog78> Maybe gcc-fortran installed the conflicting folders / files at first?

12:38 <ppetrov^> no idea

12:38 <ppetrov^> I removed gcc and installed it again with pkgadd and there were no issues. strange

12:52 <blackfrog78> The folder /usr/lib/gcc/x86_64-pc-linux-gnu/12.4.0/plugin is part of the .footprint of gcc. So it has been removed together with the port and then installed again. Did you update gcc-fortan before gcc? If yes this could indicate that gcc-fortran installed this folder dispite it being not part of its .footprint.

12:59 <ppetrov^> gcc-fortran was updated after, and it was fine

19:05 <blackfrog78> I finally managed to solve the wtmp issue for tty logins: https://pastebin.com/hrnAxJxP

19:06 <blackfrog78> With these two changes tty logins show up in the ouput of the "last" command.

20:16 <jue> blackfrog78: lastlog is deprecated since linux-pam 1.5.3, see https://github.com/linux-pam/linux-pam/releases/tag/v1.5.3. lastlog2 comes with util-linux, we need probably the wtmpdb pam-module as well

20:24 <blackfrog78> jue: I did't know lastlog has been deprecated. lastlog, at least until know, does not write wtmp records. wtmpdb looks like a good alternative.

20:24 <blackfrog78> I meant lastlog2 does not write wtmp records.

20:38 <darfo> blackfrog78: I tried your patches and last is showing root and other user ttyX activity. Thanks.