Hi Vincent
Am 30.09.24 um 18:28 schrieb Vincent Bernat:
Hey!
When a package migrates from /lib to /usr/lib, it seems that dpkg-
maintscript-helper gets confused and says that the target path (when
trying to turn a directory to symlink for example) is not owned by the
(currently installed) package, which is right.
Is there a way to make dpkg-maintscript-helper plays nice with these
changes without hacking something in each package?
Do you have a specific package which is affected by this so we can have
a look?
I suspect this is more of a special case, so would like to see a few
more details.
Am 01.10.24 um 12:17 schrieb Michael Biebl:
Am 30.09.24 um 18:53 schrieb Mark Pearson:
The firmware-sof-signed 2024.06 package:
https://packages.debian.org/trixie/firmware-sof-signed
In particular the sof-ipc4-tplg package that is being converted from a
directory to a symlink.
The dpkg-maintscript-helper complains about /usr/lib/firmware/intel/
sof-ace-tplg (and it's contents) not being owned by the package -
because they were all previously under /lib.
I suspect this is more of a special case, so would like to see a few
more details.
Holler if you need anything else.
I test upgraded firmware-sof-signed from bookworm to trixie. The
bookworm chroot was usrmerged.
Ok, I see the problem now.
The issue is, that the path used in dir_to_symlink changed pre and post usrmove. I don't think you can express that via a .maintscript file.
The attached debdiff should do.
Am 01.10.24 um 20:15 schrieb Mark Pearson:
On Tue, Oct 1, 2024, at 7:44 AM, Michael Biebl wrote:
Am 01.10.24 um 12:17 schrieb Michael Biebl:
The issue is, that the path used in dir_to_symlink changed pre and postThis looks good to me.
usrmove. I don't think you can express that via a .maintscript file.
The attached debdiff should do.
Just to confirm with an expert - I also have this MR for the issue:
https://salsa.debian.org/mpearson/firmware-sof/-/merge_requests/7/diffs?commit_id=e25d24fbbb99d8416eef2eb51bf3f1a84ce54fbe
But I think your solution is cleaner - in my limited knowledge/experience. >>
I'm going ahead and testing it, but if you have any feedback or thoughts let me know.
Thanks for the reference.
The lintian-overrides is probably a good idea. You might want to pull
that (and adapt it, regarding the comment and the the path) from that MR. That said, I prefer my solution, as it is more obvious to me what it
does and why.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 546 |
Nodes: | 16 (1 / 15) |
Uptime: | 160:20:21 |
Calls: | 10,385 |
Calls today: | 2 |
Files: | 14,056 |
Messages: | 6,416,492 |