I spent yesterday playing around with RPMS...
At this point are we sure that this bug has been fixed or are we assuming
that since a few machines worked this bug is fixed? The original test that
I did was done without opt flags at all and I got the same result. I could
post all the different combinations of opt flags I've tried and I get the
same result. With -fno_math_errno or without hasn't made any difference
At this point most all SRPMS build and test ok with my configurations with
the exception of this package.
I will test with your exact rpmrc
Tom Lane-4 wrote:
Why in the world have you got all that cruft in /etc/rpmrc? This is
just about *guaranteed* to break packages in various subtle ways,
like for instance this mysql problem you're having. If a package
doesn't build itself with -fno-math-errno, there's probably a reason.
Why not? The original test was done without any optimizations and the
same result
Anyway. After working on it all yesterday, it really looks like a bug.
> Somehow this box thought there was a 64 bit copy of gnu-smalltalk AND
> somehow uninstalled this non-existent package. My only guess is that I
> ran
> rpm --rebuilddb and a 32bit package magically became a 64bit package.
No doubt that -m64 in /etc/rpmrc had something to do with it. <----
UUm, No
But The package does not exist on koji...
http://koji.fedoraproject.org/koji/buildinfo?buildID=45706
X86_64 is excluded in the spec file of this package.
But for some reason you can yum install it.....
it's something that needs to be cleaned on Fedora's end
Thanks again,
Tony
--
View this message in context:
http://www.nabble.com/mysql-type_float.reject-tp17575369p17603936.html
Sent from the Fedora Test List mailing list archive at
Nabble.com.