I'm not 100% sure why it was removed in the first place, it looks like it got culled as part of the i686 cleanup, but multilib stuff continues to be supported by Arch.
]]>* I'm not sure about devtools not using COMPRESS* ... from what I see, when building with the devtools scripts, you're calling makepkg within the chroot and compression is being handled by /usr/share/makepkg/util/compress.sh therein. Values you specify in /etc/makepkg.conf in the chroot take precedence over the values hard-coded in that util script, so to make the change, we need to either:
1) Use a custom makepkg.conf outside of the chroot (a feature already asked for) or
2) Allow an option in the ~/.config/ccm.conf to do it automatically
* I see, you want to have the option to build passing the nocheck flag. That should be easy to add.
]]>- COMPRESSXZ
I configure it to compress multicore, had the impression ccm was slower with compresssion then i'm used to.
devtools doesn't use COMPRESS* , unless it's in a makepkg.conf that's get pulled into the chroot.
https://bbs.archlinux.org/viewtopic.php?id=245507
If ccm does support it, you should mention that somewhere.
- nocheck
https://wiki.archlinux.org/index.php/Cr … es#check()
Some of the packages I maintain/build have check() functions that fail and build is aborted.
For VCS packages this happens often, but a test failure doesn't mean the package is useless .
example
# results of llvm-check from my aur llvm-git package
Expected Passes : 29096
Expected Failures : 145
Unsupported Tests : 1203
Unexpected Failures: 58
Some functionality doesn't work, but vast majority does.
Adding the --nocheck option to makepkg commandline tells makepkg to skip the check() function and continue to package() functions.
- ccache support
- makepkg --nocheck option
Edit :
looks like the COMPRESSXZ setting from /etcmakepkg.conf isn't used ?
- I looked into ccache a while ago. At that time, I believe it wasn't support in chroots. Is that still true?
- Not sure what you mean about the --nocheck option.
- COMPRESSXZ in honored as are all others you defined including the COMPRESSZST. What problem are you experiencing with it?
There are 2 things I do miss though :
- ccache support
- makepkg --nocheck option
Edit :
looks like the COMPRESSXZ setting from /etcmakepkg.conf isn't used ?
I disabled BTRFS support until this is ressolved as I too couldn't find a reliable method to recursively delete the subvols in the proper order... what is the reasoning behind taking a snapshot anyway??
I suppose keep space usage to minimum.
]]>My workaround is to mask /usr/lib/tmpfiles.d/systemd-nspawn.conf (which is the sole cause of unasked-for subvolumes). However, there is some code in https://git.archlinux.org/devtools.git/log/?h=heftig which should find and delete subvolumes recursively.
]]>$ sudo btrfs subvolume list /
ID 454 gen 121253 top level 5 path scratch/chroot64/root
ID 455 gen 121265 top level 5 path scratch/chroot64/inglor
ID 456 gen 121255 top level 455 path scratch/chroot64/inglor/var/lib/machines
$ sudo ccm64 n
==> Nuking the chroot...
inglor@tiamat ~/cowerPkg/firefox-esr$ sudo btrfs subvolume list /
ID 455 gen 121266 top level 5 path scratch/chroot64/inglor
ID 456 gen 121255 top level 455 path scratch/chroot64/inglor/var/lib/machines
So the next time I try to build something it fails with an error such as:
$ sudo ccm64 s
----> No chroot has been created so making one now...
Create subvolume '/scratch/chroot64/root'
==> Creating install root at /scratch/chroot64/root
==> Installing packages to /scratch/chroot64/root
:: Synchronizing package databases...
core 123.8 KiB 4.03M/s 00:00 [##############################################] 100%
[..]
[log truncated]
[..]
(98/98) installing which [##############################################] 100%
:: Running post-transaction hooks...
(1/2) Updating the info directory file...
(2/2) Rebuilding certificate stores...
Initializing machine ID from random generator.
Generating locales...
en_US.UTF-8... done
de_DE.UTF-8... done
Generation complete.
----> Attempting to build package...
==> Creating clean working copy [inglor]...ERROR: cannot delete '/scratch/chroot64/inglor': Directory not empty
==> ERROR: Unable to delete subvolume /scratch/chroot64/inglor
Is that something that ccm should take care (remove the subvolumes)? Or am I using it wrongly?
Thanks
]]>Works for me.
]]>https://github.com/graysky2/clean-chroo … e/unstable
Diff against 2.68: https://github.com/graysky2/clean-chroo … ...c6b73da
Notes:
1) You need to diff ~/.config/clean-chroot-manager.conf against the new /usr/share/clean-chroot-manager/ccm.skel to make it's up-to-date.
2) Only ccm64 has been modified to date.
EDIT: Released in v2.69 live in the AUR.
]]>