Historical bug tracker for the Pacman package manager.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
FS#13051 - Allow pacman to cross-install with an --arch flag
Attached to Project:
Pacman
Opened by Aaron Griffin (phrakture) - Friday, 30 January 2009, 17:19 GMT
Last edited by Xavier (shining) - Monday, 07 September 2009, 15:06 GMT
Opened by Aaron Griffin (phrakture) - Friday, 30 January 2009, 17:19 GMT
Last edited by Xavier (shining) - Monday, 07 September 2009, 15:06 GMT
|
DetailsJust had an idea that might be nice. I am thinking from the arena of installation ISOs, here.
Allow the target architecture to be passed to pacman, using $(uname -m) if none is present. This will set a $arch variable in server parsing, similar to the $repo variable. Additionally, the functionality of util-linux-ng's setarch util may need to be used before chrooting. See http://git.kernel.org/?p=utils/util-linux-ng/util-linux-ng.git;a=blob;f=sys-utils/setarch.c Doing this allows us to very simply install a different architecture. The only thing I worry about is scriptlet execution. |
This task depends upon
Closed by Xavier (shining)
Monday, 07 September 2009, 15:06 GMT
Reason for closing: Implemented
Additional comments about closing: $arch in Server parsing is implemented. for the rest, using linux32/linux64 is fine :)
Monday, 07 September 2009, 15:06 GMT
Reason for closing: Implemented
Additional comments about closing: $arch in Server parsing is implemented. for the rest, using linux32/linux64 is fine :)
I'm attaching a sample (probably broken, not tested, not inteded for use) patch to pacman that suggests a way to get pacman to parse a $arch variable in Server entries.
If it works, it would only work when you're processing a Server entry on a machine where you want to make sure you're talking to a repo of the same architecture as the running kernel.
Attaching a patch (but it depends on my two other patches for
FS#15622)About the setarch stuff, I am unsure. It only works in some situations. For example, here is my config :
Linux xps-m1530 2.6.30-ARCH #1 SMP PREEMPT Fri Jul 31 18:10:38 UTC 2009 i686 Intel(R) Core(TM)2 Duo CPU T8100 @ 2.10GHz GenuineIntel GNU/Linux
and linux64 does not allow me to run 64 bit binaries.
Maybe if I was running a 64 bit kernel, linux32 would allow me to run 32 bit binaries. But well, it looks easy enough to run manually setarch/linux32/linux64 when we know it is supposed to work and when it is appropriate.
Also I am worried about the portability of this thing.
is it sufficient to close this ?
I also made other patches for
FS#15622, but this only checks the architecture defined in the packages.For example, if you consider this package : http://aur.archlinux.org/packages/bin32-skype/bin32-skype/PKGBUILD
it is a 32bit app made for x86_64, so it will be only installable on x86_64 (if you enable the architecture check).
If you have different needs, you can just disable the check, so don't worry.
I think what Aaron was suggesting was to move this functionality inside pacman. I first found it interesting, but after some thinking, I would much prefer to keep that code and functionality outside (in linux32 / linux64).