Skip to main content ↑ ↓
Topic: Problem with pacman. (Read 1732 times) previous topic - next topic
0 Members and 4 Guests are viewing this topic.

Re: Problem with pacman.

Reply #15 –
Regarding inkscape,

the breakage was introduced by switching from cmake to artix-cmake https://gitea.artixlinux.org/packagesI/inkscape/commit/760753a537b690cb544048a7ae455fff83cb4e44 (noticed by comparing the difference between the PKGBUILDs for the working Arch and non-working Artix). I'm not sure if where the bug lies (is Inkscape or artix-cmake at fault?), though I thought I should mention it here (I'm coming from https://gitlab.com/inkscape/inbox/-/issues/7526).

Re: Problem with pacman.

Reply #16 –
Quote from: nathanal – on
Regarding inkscape,

the breakage was introduced by switching from cmake to artix-cmake https://gitea.artixlinux.org/packagesI/inkscape/commit/760753a537b690cb544048a7ae455fff83cb4e44 (noticed by comparing the difference between the PKGBUILDs for the working Arch and non-working Artix). I'm not sure if where the bug lies (is Inkscape or artix-cmake at fault?), though I thought I should mention it here (I'm coming from https://gitlab.com/inkscape/inbox/-/issues/7526).
I'm using Artix OpenRC XFCE and I have:
Code: [Select]
jp-artix:[jp]:~$ pacman -Q cmake
cmake 3.24.2-1.1
jp-artix:[jp]:~$ pacman -Q inkscape
inkscape 1.2.1-5

Where did you find artix-cmake?
I can't even find it in the aur packages...
 

Re: Problem with pacman.

Reply #18 –
Quote from: corysanin – on
https://gitea.artixlinux.org/packagesC/cmake/src/branch/master/trunk/artix-cmake
Yes, thank you, I had seen this link, but I don't understand the need to have artix-cmake, because it is not in the repositories.

But in fact, you had indicated probable causes:
https://forum.artixlinux.org/index.php/topic,4519.msg29087.html#msg29087

Re: Problem with pacman.

Reply #19 –
Those flags that artix-cmake adds are necessary for some packages to build on the build server. Here is an example of a flag being added before artix-cmake was introduced.

No idea what's going on with inkscape, though. I don't believe artix-cmake is to blame, but I haven't looked too far into it.

Re: Problem with pacman.

Reply #20 –
Untested fix (thought I'd share before anyone spends too much time investigating): https://gitlab.com/inkscape/inkscape/-/merge_requests/4744/diffs . That at least identifies the problem, artix-cmake explicitly set CMAKE_INSTALL_LIBDIR to a absolute path, and inkscape assumed it was a relative path. 

Found due to PBS's investigation on the gitlab issue I linked earlier.

Re: Problem with pacman.

Reply #21 –
So it looks like CMAKE_INSTALL_LIBDIR is supposed to support absolute and relative paths according to the documentation. That would explain the problem though.

Re: Problem with pacman.

Reply #22 –
Quote from: Dudemanguy – on
So it looks like CMAKE_INSTALL_LIBDIR is supposed to support absolute and relative paths according to the documentation. That would explain the problem though.

Yup!

The fix has been merged into Inkscape now and the change is fairly self-explanatory (at least, combined with the cmake documentation).

I'm not familiar with the process here  :-[ . How would I get the patch (commit link, Patch link) into artix? (We can also switch back to artix-cmake, at least until the next Inkscape release).

Re: Problem with pacman.

Reply #23 –
We can just cherry-pick the commit in the PKGBUILD and rebuild. It's easy.

Edit: Just pushed an update (inkscape-1.2.1-6.1). It should work now, so you should be able to get it on the next mirror sync.

Re: Problem with pacman.

Reply #25 –
Quote from: Dudemanguy – on
We can just cherry-pick the commit in the PKGBUILD and rebuild. It's easy.

Edit: Just pushed an update (inkscape-1.2.1-6.1). It should work now, so you should be able to get it on the next mirror sync.

Awesome, thank you!

I didn't know that the using of git in PKGBUILDs. That is much better  :)