preventing md5 mismatch errors from compression changes

Toshio Kuratomi a.badger at
Tue Nov 30 04:54:44 UTC 2010

On Tue, Nov 30, 2010 at 12:15:25AM +0000, Andre Robatino wrote:
> Andre Robatino <robatino <at>> writes:
> > 2) is easy enough. To get F14 to use the new compression unconditionally, I
> > downloaded the Rawhide version of several packages and then used yum shell with
> > the commands
> > 
> > config gpgcheck 0
> > install xz-compat-libs-5.0.0-4.fc15.x86_64.rpm
> > update xz-5.0.0-4.fc15.x86_64.rpm xz-libs-5.0.0-4.fc15.x86_64.rpm
> > deltarpm-3.6-0.4.20100708git.fc15.x86_64.rpm
> > deltaiso-3.6-0.4.20100708git.fc15.x86_64.rpm
> > python-deltarpm-3.6-0.4.20100708git.fc15.x86_64.rpm
> > run
> > 
> > and to reverse the process (necessary so yum-presto will work), yum shell again
> > with the commands
> > 
> > remove xz-compat-libs
> > downgrade xz xz-libs deltarpm deltaiso python-deltarpm
> > run
> > 
> > The old and new versions of deltarpm use the old and new compression
> > unconditionally, resp.
> I'm wondering why the new xz isn't pushed to F14 and below. With xz-compat-libs
> providing, yum-presto works normally. It's only if
> deltarpm/deltaiso are updated to the F15/Rawhide versions that it breaks. Would
> anything else break - other than the fact that command-line xz would compress
> differently?
Commandline xz would compress differently and other random apps that we
don't know of would as well.  This may not cause any difficulties... OTOH,
it could be that people *are* depending on the compression being the same.
breaking that assumption is something we can do at the Fedora release
boundary... not so easily within a Fedora release.

Beyond this, there's the fact that there's not a compelling reason to update
on F14 and that the present plan for the compat libs package is that it will
go away before rawhide becomes F15.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
Url : 

More information about the devel mailing list