Hi Antonio, Thank you very much for your comments. I am learning a lot. On Sun, Sep 11, 2011 at 10:38:55AM -0700, Antonio Terceiro wrote: > You did not import a new pristine-tar entry for the 0.10.0+real version. I believe I did. I have just checked out again ruby-pdf-reader repository. There are three commits on the pristine-tar branch. Two of them mention the 0.10.0 version. The second one is in fact the same as the 0.10.0+real version. That is why there is no difference between the third and the second commit (just the tag). The third one (96dfda5...) is the (real) pristine-tar entry for 0.10.0+real. I could recover a tarball with: pristine-tar checkout ruby-pdf-reader_0.10.0+real.orig.tar.gz The reason for this +real version is that the gem did not ship the test suite. I already imported it in the repo when I realized it. I got the corresponding tarball from github, but I had to change the version, because the tag was already made public. > If you do not ship the files under bin/, there is no reason to patch them at all. You are right, I committed the patch before deciding not to ship the files in bin/. I will revert 6cb66aaa. > If you have .pc in .gitignore, it is more difficult to known why a > second build is failing. Overriding upstream's .gitignore does not seem > like a good idea, anyway. Putting .pc in .gitignore is something I got from [0]. How does one live otherwise with this uncommitted directory? (there is just one repo in pkg-ruby-extras) with a commited .pc). I thought that having an upstream .gitignore was in fact a bug in upstream release, is it not? [0] http://raphaelhertzog.com/2010/11/18/4-tips-to-maintain-a-3-0-quilt-debian-source-package-in-a-vcs/ Best wishes, Cédric
Attachment:
signature.asc
Description: Digital signature