| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| |
| |
| | |
Conflicts:
src/fedpkg.bash
|
| |
| |
| |
| |
| |
| |
| |
| | |
usage: fedpkg verify-files [-h]
Locally run 'rpmbuild -bl' to verify the spec file's %files sections. This is
useful after a successful run of 'fedpkg install' or after a 'fedpkg local'
run which failed due to %files list inaccuracies.
|
|\ \ |
|
| | | |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
usage: fedpkg.py initial-merge [-h] [-n] [repo-path [repo-path ...]]
Performs a 'git merge' of all git branches with the same content
(i.e. with the same package spec files, patch files, etc.), regardless
of their history.
This is useful after Fedora's dist-cvs to dist-git migration, as often
different branches have different histories but the same content on the
filesystem.
After these initial merges of identical trees, future merges between
the branches will be a lot easier: Easier to follow in the dependency
graph, and easier to perform without conflicts.
positional arguments:
repo-path Path to a repo to initial-merge
optional arguments:
-h, --help show this help message and exit
-n, --dry-run Whether to run without actually merging
|
|\ \ |
|
| | |
| | |
| | |
| | | |
Add man page generator to generate a man page at build time.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The definition of the --path argument to fedpkg in parse_cmdline()
already makes os.getcwd() the default value for args.path.
Thus, we can get rid of the code providing the very same default
for the path in PackageModule.__init__().
Also, make sure the path actually exists and is a directory.
|
| | |
| | |
| | |
| | |
| | | |
This avoids unnecessary lookup of non-existing users and
the resulting error messages in mock builds.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Add an option '--fix-permissions' to the 'fedpkg srpm' call.
If the '--fix-permissions' option is not given, the behaviour
of 'fedpkg srpm' does not change at all.
If the '--fix-permissions' option is given, fedpkg runs
git ls-files -z | xargs -0 chmod a+r
before running the normal rpmbuild command. This makes sure
that all 0600 files the filesystem are changed to the usual
0644 before including them in .src.rpm file.
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Unify handling of the --md5 parameter between the two targets using it:
'fedpkg local' and 'fedpkg srpm'.
If no --md5 parameter is given, then the default hashtype is used, as
determined in PackageModule.__init__() ('md5' for el5 and el6, and
'sha256' for everything else).
If an --md5 parameter is given, then use the 'md5' hashtype.
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| | |
Move the argparse setup and the actual argument parsing into
a separate parse_cmdline function.
This makes the actual main program more clear, and allows
reusing of the argparse setup for man page generation later.
|
|/ |
|
| |
|
|
|
|
|
| |
When the local branch doesn't have a merge point, catch people trying to
build from a remote url.
|
| |
|
| |
|
|
|
|
|
| |
This reverts commit 6734b0afba8cc8361392fd40b5b46a11ce0bce09.
Fixing a bad commit message
|
| |
|
| |
|
|
|
|
|
|
| |
This allows developers to force fedpkg into not discovering the
distribution based on branch name, and instead just take what the
developer says it is.
|
| |
|
| |
|
|
|
|
|
| |
This looks to see what tag 'dist-rawhide' will build into, which will
tell us good info about the dist values.
|
|
|
|
| |
This will be useful to query koji without bothering to log in
|
| |
|
|
|
|
|
|
| |
This reverts commit e50d2677e6d9ab661b791097dde3086ecf21c16f.
Better to just download sources here, not make the srpm.
|
|
|
|
|
|
|
| |
Compare the modification times of the spec and the srpm if it exists and
see if we need to re-generate due to a spec change. Most of this code
comes from Gianluca Sforna <giallu@gmail.com> with modifications from
me.
|
| |
|
|
|
|
|
| |
We can't trust the name of the specfile to match, so get the name as
defined within the specfile.
|
| |
|
|
|
|
|
| |
This required a new option to mockbuild, md5. This is because srpm
needs this argument to function correctly.
|
|
|
|
|
| |
There are just too many error types here that can trip us up, so use the
dreaded catchall.
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Patch from Toshio
https://bugzilla.redhat.com/show_bug.cgi?id=667433#c7
|
|
|
|
|
| |
patch from Toshio
https://bugzilla.redhat.com/show_bug.cgi?id=667433#c6
|
|
|
|
| |
This can also prevent auth errors (#634946)
|
|
|
|
| |
This comes from Brian Lane, ./git-changelog -v <tag>
|
| |
|
| |
|
| |
|
|
|
|
| |
This fixes a potential traceback on ^c
|
|
|
|
|
| |
Looks like argparse doesn't like a % in help. Odd. Also fix a logic
error for non-tty usage.
|
|\
| |
| |
| |
| | |
Conflicts:
src/pyfedpkg/__init__.py
|
| |
| |
| |
| |
| |
| | |
When the "sources" target is run, then curl prints some statistics during
the download of the missing file, but it doesn't print the filename that gets
downloaded.
|
|\ \ |
|