[Bug 529548] Review Request: mingw32-libogg - MinGW build of the libogg bitstream library
bugzilla at redhat.com
bugzilla at redhat.com
Mon Oct 19 13:34:37 UTC 2009
Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.
https://bugzilla.redhat.com/show_bug.cgi?id=529548
--- Comment #4 from Richard W.M. Jones <rjones at redhat.com> 2009-10-19 09:34:36 EDT ---
(In reply to comment #3)
> > That's simply not supported. Koji might happen to run on (eg) x86-64
> > or PPC builder, where Wine either can't be installed or couldn't run even
> > if it could be installed. Don't worry about it.
I should explain that there are three related issues here which
stop us from running Wine. Two are bugs (IMHO) in RPM
or Koji. One is a bug (IMHO) in Wine itself.
(1) RPM/Koji doesn't understand cross-compilation at all. Our
target packages are all noarch which is (sort of) correct within
the bounds of RPM[*]. Because the package is noarch, Koji decides
that it can be built on any machine, so it can choose a PPC
builder for example. On PPC, the cross-compiler works, but Wine
could never run. But there is no way to say to Koji that a
package is noarch, but should be built anyway on i386.
[*] If you prefer you can think of wine as an interpreter,
interpreting Win32 PE executables, something like:
foo.pl is interpreted by /usr/bin/perl
foo.exe is interpreted by /usr/bin/wine
(2) Koji won't install Wine on x86-64, because Wine is an i386
package. Koji doesn't understand multilib (probably a good
thing, because multilib is crack).
(3) Wine after installation doesn't "just work". You have to
fiddle with a difficult configuration file:
http://fedoraproject.org/wiki/MinGW/Configure_wine
which is difficult to automate. Wine used to have an
environment variable (WINE_PATH or something?) but they
removed it(!)
> Thanks, I thought I was missing something. I'm still running all tests I can
> manually - e.g., working on mingw32-xz now, and some edge case tests fail
> depending on the MINGW32_CFLAGS you pick, which would seem to point to some
> compiler problems, so properly passing all tests should be a priority.
As long as you're testing it, that's good. It doesn't look like
we'll ever get the Koji issues fixed.
--
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
More information about the package-review
mailing list