Am 03.01.2011 15:46, schrieb Dan HorĂ¡k:
here are few of my thoughts before starting koji-shadow
- need to ask dgilmore for the package list that are expected to be
missing on ppc/ppc64 (ExcludeArch/ExclusiveArch), he has a script for
that
- because the difference between primary and ppc is so large and there
were mass rebuilds of all python (for 2.7) and perl (for 5.12) packages,
we shouldn't use the "import-noarch" feature, too new packages would be
available in the buildroots causing broken dependencies (the
"prefer-new" feature is needed to include fixed builds)
- the injected packages, that do not exist in primary koji, should have
<your_initials>.X suffix in the Release field, so they are later easily
identified
- should we really start koji-shadow against dist-f15 or should we start
with dist-f14 first even without an intention to be complete, just to
shorten the gap before starting dist-f15
Additions, opinions?
Dan
python and perl will be a big mess anyway, I think we should update
them, import the noarch packages afterwards and fix what whatever we
broke by doing it that way.
+100 to the suggestion that package with some patches that don't exist
in primary koji should have your initials in the release field.
We've done that on s390x and it helped a lot to keep track of those
packages until the patches got into the packages on the primary archs.
I'd rather concentrate on F15 than filling the web-interface with F14
failures. For some packages it might be worth building the F14 package
requirements, but that should be done depending on the problem, not for
all packages.
Karsten