What I have been doing with my patches is to keep all the revisions in JIRA but try to name such that (hopefully) it is easier for the reviewers to see what is the latest patch. The naming convention I use has bug number, little info about bug and date it was created. for eg
Derby479LinkageErrorReturnNullIfNulldiff021306.txt
Mamta
On 2/16/06, Satheesh Bandaram <[EMAIL PROTECTED]> wrote:
Daniel John Debrunner wrote:I can see arguments either way... Last time I thought this was discussed, the consensus was to delete previous patch/spec if the new one completely replaces the old one. Here is the link to previous discussion.I notice that some folks are deleting jira attachements when they attach a new version of a patch or functional/design spec. I personally think this should not be encouraged, it's sometimes helpful to have the history of the patch and especially helpful to have the history of a spec.Does anyone think it's a good idea to delete old attachements? Dan.
http://www.nabble.com/what-is-minimum-correct-protocol-for-posting-patch-to-list--t65135.html#a177731
Part of the message from there:
Mike Matrigali wrote:
> what is the minimum correct protocol when submitting patches:
> attach to jira
> attach to mail on list
> send in line on list
>
....
DERBY-332 is a poster child for confusion. I applied a patch on May 5
that was submitted on May 4. The issue was opened for it on June 1 and
the April 27 patch uploaded. In the meantime I had even forgotten that I
had applied a patch back in May. It got untangled yesterday.
If I remember right, it was just a month ago that we were discovering
that we had so many patches posted to derby-dev that we were starting to
lose track of them, especially when new patches were replacing old ones.
If it's in Jira and a new patch supercedes an old patch, it's easy to
remove that old patch so it doesn't get applied by mistake. Or if the
old patch should be left for some reason, it's easy to see that there
are multiple patches and judge which one should be applied.
-jean
