I'm getting digest errors on lapack.tgz, causing both lapack-atlas and lapack-reference to fail. Moreover, lapack-20020531-20021004.patch.bz2 now fails to apply. Is there any reason we don't keep a versioned tarball on our mirrors?
Created attachment 107380 [details] patch failure log Adding the patch failure log.
Probably nobody's gotten around to it. Want to make the change?
I would, but doing it right really requires generating a new patch, which is going to take more time than I have right now.
(In reply to comment #3) > I would, but doing it right really requires generating a new patch, which is > going to take more time than I have right now. Nah, just take the old tarball, version it and upload it, then change the SRC_URIs as needed. We'll need a revbump for the new lapack + new patch.
The new lapack has been on my to-do list for a long time now. Maybe I get to it this weekend. Unfortunately, I am currently hunting for a new apartment so I suspect I'll be somewhat short on time. Markus
(In reply to comment #4) > Nah, just take the old tarball, version it and upload it, then change the > SRC_URIs as needed. We'll need a revbump for the new lapack + new patch. Ah, I had discarded that option because I don't happen to have an old tarball. Do you have one, per chance?
The one on the mirrors should still be the old one unless someone has secretly bumped the ebuild ;) cheers, Markus
Upstream has renamed lapack.tgz to lapack-3.0.tgz to accommodate their most recent lapack-3.1 release; otherwise the tarballs are identical. I've changed all lapack-reference and lapack-atlas ebuilds accordingly and also pushed a lapack-3.0.tgz tarball out to the mirrors. In doing so I noticed that we need to stabilize lapack-reference-3.0-r4 since the current stable lapack-reference ebuild doesn't work with gfortran. Best, Markus