You mention committing a patch to the current trunk.
Which makes me wonder, what should be the process for providing patches for only a branch if something doesn't apply to the trunk? No different? Or is this too rare to document?
And will a committer automatically apply a patch to another branch if the 'affects version' also indicate that branch? Or does that require a separate patch? (I think the merge commands have been submitted by contributors, not sure if that was an actual decided-on process).
Myrna
On 2/23/06, Bryan Pendleton <[EMAIL PROTECTED]> wrote:
> 3) There is some special svn processing needed for added/deleted files
> in patches. Perhaps some explanation of that would be good or a link to
> someplace that does.
Yes, I'd like to incorporate this content. But I'm not quite sure what
it should say. I believe you're talking about the problem that is raised
in the mail thread here: http://svn.haxx.se/dev/archive-2004-09/0469.shtml
Unfortunately, that page doesn't really explain the answer, because I
think that maybe there isn't a perfect answer right now.
Here's what I think I might put into the db-derby Wiki page. What do
you think?
Note that the patch format as produced by "svn diff" is not perfect;
in particular, it does not clearly describe changes which entirely
delete existing files or which add brand-new files. For more background
on the problem, read: http://svn.haxx.se/dev/archive-2004-09/0469.shtml
If the patch adds or deletes a file, this should have been mentioned
explicitly by the developer when the patch was attached to the JIRA
issue, and the output of "svn stat" should also have been attached to
assist the committer and reviewers.
Here's how to apply a patch which includes adds and/or deletes:
* apply the patch normally, using "patch -p0"
* read the "svn stat" output from the JIRA issue, and take note of
each file which is to be (A)dded or (D)eleted.
* For each file, "svn revert" the file to undo the changes made by
patch, then issue "svn add" or "svn delete" as appropriate.
* Issue "svn stat" on your sandbox, and compare the results against
the output attached to the JIRA issue.
thanks,
bryan
