udisks2 build error #18
Loading…
x
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
I have tried building udisks2 with lvm2 and btrfs-progs and without.
I get this error log without btrfs-progs installed.
https://pastebin.com/vnrfq9Mm
I get this error with btrfs-progs installed.
https://pastebin.com/bMSVbBXS
I normally do not build btrfs into my kernels. However I did try adding it to my kernel but still no luck.
plasma-meta installation pulled in this package just an fyi. From what I can tell, doxygen package does not exist except in ppetrov or stenur repos. Not sure if thats my issue.
You need to rebuild libblockdev with btrfs support due to a recent update to the libblockdev Pkgfile:
499c4b858c
But it also fails because it wants libblockdev to have parted support:
which can be fixed by not building the udisks daemon by default.
That I can commit later.
You can try:
and see if that fixes the problem you reported.
Working down the list, seems it also wants parted as you mentioned, but then it wants libblockdev-nvme.
Would it be better to make parted, btrfs-progs and lvm2 part of the depends instead of optional? Or depends in libblockdev?
libblockdev requires btrfs-progs, lvm2, parted and nvme-cli to build correctly when trying to pull in plasma-meta.
Thats a pretty common problem: trying to make things as small as possible (from the dependency chain view) and making them compatible.
A wayland desktop requires you to rebuild mesa among a few other ports to work correctly, too. Right now there is no way to do that better.
Take your fastfetch Pkgfile for example: https://pastebin.com/8LFWNPFq
You don't need chafa, libdrm, imagemagic: https://github.com/fastfetch-cli/fastfetch/blob/dev/CMakeLists.txt#L47
Those are options you can control, and thats what I try to aim for generally. This method can produces problems, but they can also be worked around, e.g. with libreoffice I have https://git.crux.nu/ports/contrib/src/branch/3.7/libreoffice/Pkgfile#L53 in there.
As kde-plasma-6/solid is the port that pulls in udisks2, I'll see if there is anything we can change there to not depend on these features by default.
Edit: For me, solid builds just fine with udisks2 and libblockdev as is though. Can you share your build logs?
Yes, I understand some of the options for fastfetch are optional. However, it's a personal repo and not published. Thats ok of course and it does make for a good example. I am aware of the KISS principle for Crux and to not include extras if needed as well.
I mean, if I could have a basic fresh install of Crux in a VM and then type in prt-get depinst udisks2 I would expect udisks2 to compile exactly as intended. This is where I falter... Lots of packages have nuances even if the deps are not listed and can use programs in the system that may not be part of the dependencies. So, I often do full recompiles every so often to hopefully evade some of the pitfalls. While possibly also gaining some.
udisks2 - successful build
https://pastebin.com/u3uSA7s3
libblockdev - successful build
https://pastebin.com/mWg6SXn0
soild - successful build
https://pastebin.com/AKBe9CtZ
After I had btrfs-progs installed the results changed as posted earlier, needing parted. But I was still getting the nvme error and that required me to install nvme-cli and it's dependency libnvme. Unfortunately upon removing nvme-cli and then recompiling libblockdev I am no longer getting the nvme error from before. So I cannot reproduce the log with the error.
That was all I was trying to show here. Of course its fine, if thats how you want to use it.
fwiw, I build in a container thats pretty much as near to a vanilla core install as can be. So udisks2 (with the recent changes) will build just fine.
Since your logs all build successfully, I don't think there is anything else to do here.