- 01 Aug, 2013 3 commits
-
-
Natanael Copa authored
seems like configure thinks we are crosscompiling unless we set CC when --build and --host are specified.
-
Natanael Copa authored
-
Natanael Copa authored
-
- 30 Jul, 2013 2 commits
-
-
Natanael Copa authored
this is to catch bugs in abuild itself early
-
Natanael Copa authored
it wrongly scanned /lib instead of $datadir/lib, causing bad things to happen when scanning provides
-
- 29 Jul, 2013 4 commits
-
-
Timo Teräs authored
its only arm that needs it
-
Natanael Copa authored
instead let apkbuilds use --host and --build
-
Timo Teräs authored
and exit with error if not
-
Natanael Copa authored
This reverts commit 4234a10e. We will not use config.guess. Instead we will use --build and --host everywhere.
-
- 26 Jul, 2013 5 commits
-
-
Natanael Copa authored
instead, provide 'update_config_sub' and let aports that needs replace config.sub call it.
-
Natanael Copa authored
-
Natanael Copa authored
we want be able to run build and rootpkg separately. so you can stop in the process, fix things, and then continue with 'abuild rootpkg' to generate the package manually.
-
Natanael Copa authored
If some package fails, then stop processing the rest. This is so we don't risk test build the rest for the packages against wrong lib. For example, if 'abump libfoo-2.0 foo-2.0' fails on libfoo, we don't want testbuild foo-2.0, which might end with success against libfoo-1. We also echo the list of packages that was not bumped, so its easy to copy/paste next run.
-
Natanael Copa authored
-
- 25 Jul, 2013 1 commit
-
-
Natanael Copa authored
-
- 23 Jul, 2013 1 commit
-
-
Natanael Copa authored
Use a global var for last_commit so we dont need call git for every subpackage
-
- 22 Jul, 2013 3 commits
-
-
Natanael Copa authored
For example, to mask out musl, add: options="!libc_musl"
-
Natanael Copa authored
We need set it in global scope as we need it even if we have not called sanitycheck
-
Natanael Copa authored
-
- 19 Jul, 2013 4 commits
-
-
Natanael Copa authored
we better fix the affected apkbuilds
-
Natanael Copa authored
-
Timo Teräs authored
-
Timo Teräs authored
-
- 16 Jul, 2013 1 commit
-
-
Natanael Copa authored
-
- 12 Jul, 2013 2 commits
-
-
Natanael Copa authored
-
Natanael Copa authored
-
- 10 Jul, 2013 3 commits
-
-
Natanael Copa authored
It appears that when the subshell has a ||, the 'set -e' within subshell gets invalidated. This will work as expected: ( set -e; false; echo "should not get here" ) While this will not work as expected: ( set -e; false; echo "should not get here" ) || false We resolve it by using $? to detect the status of subshell. We also let the exitcode indicate how many packages that failed. While here we also refactor it so most of the loop happens within the subshell. This lets us set (or increase) rc variable once, and it reduces number of forks which gives slightly better performance.
-
Natanael Copa authored
We actually want be able to re-run abump without needing reset the pkgver in case we had to fix things. This reverts commit 8198ded8. Conflicts: abump.in
-
Natanael Copa authored
-
- 09 Jul, 2013 11 commits
-
-
Natanael Copa authored
-
Natanael Copa authored
Otherwise abuild-sign will end with error
-
Dubiousjim authored
This permits introducing branch-specific modifications to abuild.conf settings. Any settings there override /etc/abuild.conf and $HOME/.abuild/abuild.conf.
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-
Dubiousjim authored
-