buildsys September 2014

buildsys@lists.fedoraproject.org

13 participants
7 discussions

[PATCH] Allow spec file to be in a different path than /
by Thomas 24 Oct '14

24 Oct '14

This should allow people to build from git repositories that have a different structure than Fedora (they still need some magic with source_cmd). The patch is not too intrusive and can be set per repository. Let me know if it can be considered for inclusion. I am happy to rework it. --- diff --git a/builder/kojid b/builder/kojid index b3d0d1c..89f9474 100755 --- a/builder/kojid +++ b/builder/kojid @@ -3292,7 +3292,7 @@ class BuildSRPMFromSCMTask(BaseBuildTask): self.patch_scm_source(sourcedir, logfile, opts) # Find and verify that there is only one spec file. - spec_files = glob.glob("%s/*.spec" % sourcedir) + spec_files = glob.glob("%s/%s*.spec" % (sourcedir, scm.spec_path)) if len(spec_files) == 0: raise koji.BuildError("No spec file found") elif len(spec_files) > 1: diff --git a/koji/daemon.py b/koji/daemon.py index 205f62e..68af212 100644 --- a/koji/daemon.py +++ b/koji/daemon.py @@ -200,6 +200,7 @@ class SCM(object): - use_common (defaults to True, may be set by assert_allowed()) - source_cmd (defaults to ['make', 'sources'], may be set by assert_allowed()) - scmtype + - spec_path The exact format of each attribute is SCM-specific, but the structure of the url must conform to the template above, or an error will be raised. @@ -220,6 +221,7 @@ class SCM(object): self.revision = fragment self.use_common = True self.source_cmd = ['make', 'sources'] + self.spec_path = '/' for scmtype, schemes in SCM.types.items(): if self.scheme in schemes: @@ -311,6 +313,10 @@ class SCM(object): else: # there was nothing after the trailing :, so they don't want to run a source_cmd at all self.source_cmd = None + # check if spec path is defined + if len(scm_tuple) >= 5: + if scm_tuple[4]: + self.spec_path = scm_tuple[4] + '/' break else: self.logger.warn('Ignoring incorrectly formatted SCM host:repository: %s' % allowed_scm)

3 6

[PATCH] Added a brute-force command to do a submodule-update after branch-checkout.
by Martin Wilhelm 08 Oct '14

08 Oct '14

Reason: 1) We use mock to build a RPM. Mock does all the SCM-stuff via its plugin. 2) This plugin actually does NOT do a "git submodule update --init --recursive" after the branch checkout. 3) This leads to wrong (old) submodule references. 4) Compile error. --- py/mockbuild/scm.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/py/mockbuild/scm.py b/py/mockbuild/scm.py index e76293d..dc67bf0 100644 --- a/py/mockbuild/scm.py +++ b/py/mockbuild/scm.py @@ -39,6 +39,7 @@ class scmWorker(object): self.branch = None self.postget = None + self.afterpostget = None if 'branch' in opts: self.branch = opts['branch'] if self.branch: @@ -46,6 +47,7 @@ class scmWorker(object): self.get = self.get.replace("SCM_BRN", "-r " + self.branch) elif self.method == "git": self.postget = "git checkout " + self.branch + self.afterpostget = "git submodule update --init --recursive" elif self.method == "svn": self.get = self.get.replace("SCM_BRN", self.branch) else: @@ -73,6 +75,7 @@ class scmWorker(object): self.log.debug("SCM checkout command: " + self.get) self.log.debug("SCM checkout post command: " + str(self.postget)) + self.log.debug("SCM checkout after post command: " + str(self.afterpostget)) self.environ = os.environ.copy() # Set HOME properly while checking out from SCM since tools like # Subversion might have there settings needed to carry out checkout @@ -90,6 +93,8 @@ class scmWorker(object): mockbuild.util.do(shlex.split(self.get), shell=False, cwd=self.wrk_dir, env=self.environ) if self.postget: mockbuild.util.do(shlex.split(self.postget), shell=False, cwd=self.src_dir, env=self.environ) + if self.afterpostget: + mockbuild.util.do(shlex.split(self.afterpostget), shell=False, cwd=self.src_dir, env=self.environ) self.log.debug("Fetched sources from SCM") decorate(traceLog()) -- 1.9.1

3 3

[mock PATCH] drop the devel link for ppc
by Dan Horák 30 Sep '14

30 Sep '14

There is no rawhide for ppc built since F-21. Signed-off-by: Dan Horák <dan(a)danny.cz> --- mock.spec.in | 1 - 1 file changed, 1 deletion(-) diff --git a/mock.spec.in b/mock.spec.in index 24f3dd5..4de572e 100644 --- a/mock.spec.in +++ b/mock.spec.in @@ -66,7 +66,6 @@ ln -s epel-5-x86_64.cfg fedora-5-x86_64-epel.cfg # more compat, from devel/rawhide rename ln -s fedora-rawhide-i386.cfg fedora-devel-i386.cfg ln -s fedora-rawhide-x86_64.cfg fedora-devel-x86_64.cfg -ln -s fedora-rawhide-ppc.cfg fedora-devel-ppc.cfg ln -s fedora-rawhide-ppc64.cfg fedora-devel-ppc64.cfg popd echo "%defattr(0644, root, mock)" > %{name}.cfgs -- 1.9.3

3 3

miscellaneous patches for rpkg
by Lars R. Damerow 26 Sep '14

26 Sep '14

Hi all, I wrote a wrapper for rpkg to assist in our internal RPM builds, and while doing so I fixed a few bugs and added some minor features. Might someone review them and consider them for inclusion upstream? thanks! -lars

4 13

[PATCH] sanity check requested arches for image builds
by Mike McLean 24 Sep '14

24 Sep '14

3 2

docker-rpm-builder v0.9 released
by Alan Franzoni 19 Sep '14

19 Sep '14

Hello, after the proof of concept I posted about some months ago I elaborated a little on the topic. Pre-release version 0.9 can be found here: https://github.com/alanfranz/docker-rpm-builder/archive/0.9.tar.gz What does it do? It uses docker to build a native rpm from a source directory holding a .spec and source/patch files; it's VERY useful in CI environments, wherever you need to build an RPM as your final product. At the moment is still a bit rough, but it's very small and it should be very easy to make it fit to anyone's needs. Advantages over plain mock: - work on any host distribution as long as it supports docker; - doesn't require the creation of a source rpm before building a binary - as far as I have been testing, it's much faster. If anybody tries it and thinks it's worth contributing, contact me and/or open tickets or pull request on github. Bye, -- My development blog: ollivander.franzoni.eu . @franzeur on Twitter contact me at public(a)[mysurname].eu

2 3

koji mergerepo error?
by Pat Riehecky 08 Sep '14

08 Sep '14

Just installed my first EL7 koji slave and merge repo is kicking out this error: Traceback (most recent call last): File "/usr/libexec/kojid/mergerepos", line 276, in <module> main(sys.argv[1:]) File "/usr/libexec/kojid/mergerepos", line 271, in main merge.write_metadata() File "/usr/libexec/kojid/mergerepos", line 251, in write_metadata mdgen.doPkgMetadata() File "/usr/lib/python2.7/site-packages/createrepo/__init__.py", line 404, in doPkgMetadata if '://' in pkg: # remote File "/usr/lib/python2.7/site-packages/yum/packages.py", line 384, in __getitem__ return getattr(self, key) TypeError: getattr(): attribute name must be string Figured I'd pass it along.... -- Pat Riehecky Scientific Linux developer http://www.scientificlinux.org/

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

buildsys September 2014