On Thu, 28 Feb 2013 10:15:35 -0700, Brad Bell wrote:
and would appreciate any feed back or comments about them.
In the past I have made the mistake of running
In cases that failed and could have been avoided using scratch builds; see
Why do you consider it a mistake?
If the "fedpkg build" succeeded, you would save yourself the previous
scratch-build. If the "fedpkg build" failed, why would that be a problem?
You could fix the package and resubmit the build (even with the same EVR).
It now seems to me preferable to do the following before uploading
wget http:///path/to/yournewtarball.tar.gz .
rmpbuild -bs <package_name>.spec
koji build --scratch rawhide <package_name>.srpm
What does the "preferable" refer to? And what is the wget command used
for? To retrieve source tarballs specified inside a spec file, you can
use "spectool -g NAME.spec". To automate building the src.rpm from
within Fedora package git, you could even use "fedpkg srpm".
Of course, you are free to prepare your update outside the Fedora pkg git
tree. That means, run "fedpkg srpm" (or "git pull ; fedpkg srpm"),
the resulting src.rpm to your local rpmbuild tree, update/upgrade/modify
it there, run a Mock build, and finally copy the changed/new files into
the git tree and check them in.
The three-line example you've given above is a very shortened version of a
package upgrade. Typically one would examine the downloaded tarball prior
to using it, and one would prepare and evaluate a local Mock build or a
scratch-build before deciding whether to publish the update via Fedora.
Fedora release 18 (Spherical Cow) - Linux 3.7.9-201.fc18.x86_64
loadavg: 0.00 0.04 0.11