Advice on managing Git, Branches & Patches
Hello All, I have some more questions about git. Really, what I'm interested in is how everyone handles fixing bugs and adding new features to koha, and submitting the patches. At the moment I have a private git repo that is up to date with git.koha.org's master branch. I have 5 commits on top of it, representing five new features. I was copying my master branch, adding a feature, then merging back into master, but some strange things were happening. It should have been a straight fast-forward, nothing had changed in master, but instead it seems as if the new commit was being merged in between other commits I had made, does anyone know what happened? To make five patches to submit, I created five branches based on master, then rebased each one leaving only one of the feature commits for each, thus ending up with five patches starting in 0001. Is this a reasonable way to work, or is there a better solution? Also, I submitted all five patches at once, with git send-email, but only four made it through. The 5th patch ended up bouncing back to me with the message CORRUPTED MESSAGE. This has happened to me before, but I found it very odd that it would work fine for 4 out of 5 patches, all submitted with one command. Thanks for the advice, Kyle http://www.kylehall.info Information Technology Crawford County Federated Library System ( http://www.ccfls.org )
On Wed, Jan 21, 2009 at 9:52 AM, Kyle Hall <kyle.m.hall@gmail.com> wrote:
To make five patches to submit, I created five branches based on master, then rebased each one leaving only one of the feature commits for each, thus ending up with five patches starting in 0001. Is this a reasonable way to work, or is there a better solution?
Yep, that works fine. The number doesn't really matter to the receiver.
Also, I submitted all five patches at once, with git send-email, but only four made it through. The 5th patch ended up bouncing back to me with the message CORRUPTED MESSAGE. This has happened to me before, but I found it very odd that it would work fine for 4 out of 5 patches, all submitted with one command.
That's odd. The only time I have seen that is when the patch contains *very* long lines, such that git send-email chokes on it. In that case, the solution is to communicate the patch via alternate means like posting to a webpage, gitweb, FTP, etc. It may be helpful to update your version of git if not current, also. Glad to see your contributions hitting the patches list! --Joe
participants (2)
-
Joe Atzberger -
Kyle Hall