Kristian Waagan wrote:
On 03.05.10 21:36, Rick Hillegas wrote:
Hi Kristian,
Some comments inline...
Kristian Waagan wrote:
Hi,
When testing the source bundle(s), I noticed:
a) the contents .zip and the tar.gz are equal (a good thing)
Hooray!
b) we include the maven directory, but not the maven2 directory
(see DERBY-4632, the Maven 1 files are suggested deleted)
Should we include maven2 and exclude maven, or remove it
Thanks for pointing this out. I have corrected the release-building
targets to include maven2 but not the original maven directory which you
are going to deprecate.
c) we don't include any IDE files/projects. Is this intentional?
Thanks to you and Knut for drawing my attention to DERBY-3857. I also
corrected the release building targets to pull in the ide subtree.
Thanks, Rick.
A next step here (independent of what you improved now), is to make
sure the IDE project files are in a good state. Maybe we get more
feedback on this now that we're including the files in the src bundle.
I notice that DERBY-3857 was open for a long time even though the fix
was trivial. I don't think that the category "build tools" is specific
enough to get the attention of our release managers. Maybe we need to
flag these issues with a keyword to distinguish release-generation
issues from ordinary defects in our cumbersome build.
+1
Is this new keyword already documented in the release process wiki page?
Thanks for the reminder. I have added another step, mentioning this tag,
to the following section of the release instructions:
http://wiki.apache.org/db-derby/DerbySnapshotOrRelease#Work_the_toward_a_release_candidate
Thanks,
-Rick
I also tried building Derby on various system, and except for some
build automagic issues and an ant issue, this worked fine using the
source bundle.
I don't feel that these defects warrant a re-spin by themselves.
However, I merged the release-building changes to the 10.6 branch in
case we need another candidate.
I completely agree, I'm just working my way through issues I
discovered during testing of the RC.
I do however observe some hangs (client/server), but only a few tests
seem to be affected on a few specific platforms / JVMs (I'll add more
information soon).
Thanks,
