It’s not true that “we cannot test the backcompatibility of the 6.5.0 branch 
with itself”.  After releasing a *.0 releae, the RM can just set the release 
branch version to *.1, and then there are no issues with adding the *.0 
backcompat indexes.

I believe the real reason these don’t get added to the release branches is 
economy of effort.  It’s not certain that there will be a *.1 release after a 
*.0 release, so why bother?

This is a constant source of confusion, though.  Effort is definitely not 
economized when considering an RM who’s never done a bugfix release before.

Some perspective: 8/12 of the 5.X and 6.X relase branches had, or will have 
(6.5.1), at least one bugfix release.  It’s now the *ordinary* case that 
release branches will get a bugfix release.

I think it’s time to change the ReleaseToDo to tell RMs to always generate the 
backcompat indexes on the release branch, regardless of whether the current 
release is a bugfix release.

--
Steve
www.lucidworks.com

> On Apr 9, 2017, at 6:05 PM, jim ferenczi <[email protected]> wrote:
> 
> Ok sorry I should have been more specific. The backcompat tests are not 
> created on the release branch for the first minor release (eg. 6.5.0). They 
> are only created for the master branch and the 6x branch. Then during the 
> first bugfix of the current release branch (eg. 6.5.1) we push the backcompat 
> test directly on the release branch. This is not done before because we 
> cannot test the backcompatibitily of the 6.5.0 branch with itself. 
> 
> 2017-04-09 22:57 GMT+02:00 Joel Bernstein <[email protected]>:
> Thanks Jim, I don't quite understand the rational for when the backcompat 
> indexes are created, but that's OK. I'll create a new RC this evening. 
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Sun, Apr 9, 2017 at 4:44 PM, jim ferenczi <[email protected]> wrote:
> Joel,
> The backcompat indexes are not added for a minor release. They are added on 
> the first bugfix release on the minor branch. There is a note in the TODO:
> "Make sure that the backcompat index for the previous release has been added 
> to the release branch. (Note that this will ordinarily not have been done if 
> the current release is X.Y.1, i.e. the first bugfix release off the stable 
> branch.) See the post-release section "Generate Backcompat Indexes" below - 
> remember you'll be generating an index for the previous release."
> 
> I just pushed the backcompat indices in the release branch. You'll need to 
> generate a new release candidate though.
> 
> 2017-04-09 3:15 GMT+02:00 Ishan Chattopadhyaya <[email protected]>:
> No, this has not changed. I think backcompat indexes for the previous release 
> was not added. The 6.5.0 's RM might've missed this step.
> 
> On Sun, Apr 9, 2017 at 4:45 AM, Joel Bernstein <[email protected]> wrote:
> Looks like I need to add the back compat indexes. In  the releaseTodo this is 
> post release activity but it looks that has changed.
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Sat, Apr 8, 2017 at 6:58 PM, Joel Bernstein <[email protected]> wrote:
> I don't believe I've missed any steps listed:
> https://wiki.apache.org/lucene-java/ReleaseTodo
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Sat, Apr 8, 2017 at 6:56 PM, Joel Bernstein <[email protected]> wrote:
> Ok, the keys appear to be sorted out now. Smoke test now gets much further 
> but fails with the error below. I'll go back see if I've missed a step...
> Releases that don't seem to be tested:
> 
>   6.5.0
> 
> Traceback (most recent call last):
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 1476, in <module>
> 
>     main()
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 1420, in main
> 
>     smokeTest(c.java, c.url, c.revision, c.version, c.tmp_dir, c.is_signed, ' 
> '.join(c.test_args))
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 1458, in smokeTest
> 
>     unpackAndVerify(java, 'lucene', tmpDir, 'lucene-%s-src.tgz' % version, 
> gitRevision, version, testArgs, baseURL)
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 622, in unpackAndVerify
> 
>     verifyUnpacked(java, project, artifact, unpackPath, gitRevision, version, 
> testArgs, tmpDir, baseURL)
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 768, in verifyUnpacked
> 
>     confirmAllReleasesAreTestedForBackCompat(version, unpackPath)
> 
>   File "dev-tools/scripts/smokeTestRelease.py", line 1396, in 
> confirmAllReleasesAreTestedForBackCompat
> 
>     raise RuntimeError('some releases are not tested by 
> TestBackwardsCompatibility?')
> 
> RuntimeError: some releases are not tested by TestBackwardsCompatibility?
> 
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Sat, Apr 8, 2017 at 2:53 PM, Joel Bernstein <[email protected]> wrote:
> My key has appeared: http://home.apache.org/keys/group/lucene.asc.
> 
> I'll work on an RC this evening.
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Fri, Apr 7, 2017 at 5:33 PM, Joel Bernstein <[email protected]> wrote:
> Ok, I've added the PGP fingerprint to my account on id.apache.org. I'll wait 
> until step #1 completes. 
> 
> Then I'll populate the three key files mentioned in Ishan's notes. 
> 
> Then I'll regenerate the RC.
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Fri, Apr 7, 2017 at 5:20 PM, Joel Bernstein <[email protected]> wrote:
> I need to get me public key into my profile on id.apache.org. I'll work on 
> that first.
> 
> Joel Bernstein
> http://joelsolr.blogspot.com/
> 
> On Fri, Apr 7, 2017 at 4:55 PM, Steve Rowe <[email protected]> wrote:
> Joel,
> 
> 
> > On Apr 7, 2017, at 4:36 PM, Steve Rowe <[email protected]> wrote:
> >
> > a key generated with gpg2 won’t be visible to gpg.
> 
> Lower-impact fix (maybe) than symlinking - this will make your public key 
> visible to ‘gpg’:
> 
> $ gpg --recv-key EE64CB1E
> 
> --
> Steve
> www.lucidworks.com
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to