Release Trac 0.11.5

[cboos]

This task ticket is used to coordinate the finalization and testing of the next stable version of Trac, 0.11.5.

Finalization

Besides the usual steps listed in TracDev/ReleaseChecklist, we want to ensure that the next release of Trac is as bug free as possible, so we will soon freeze development of the 0.11-stable branch (estimated 2009-06-27), provide a first release candidate package (0.11.5rc1) and start a one week testing period (estimated 2009-06-28 to 2009-07-04). If no critical problem is found during that period, the effective 0.11.5 release can be made. If a problem is found, the cycle repeats with a new release candidate and one week (or less if the fix is trivial) testing period.

Testing

People having volunteered to contribute to release testing were added to the CC: list.

If you'd like to contribute, please fill an entry in the table you'll find there, detailing the configuration(s) you will test, and report the status of your testing in this ticket.

[cboos]

I started to write the ChangeLog#a0.11.5.

[cboos]

Testing 2.3.5 with MySqlDb:

2009-06-29 17:18:21,454 Trac[loader] ERROR: Skipping "trac.db.mysql = trac.db.mysql_backend": (can't import "ImportError: cannot import name Popen")

Testing 2.3.5 with PostgreSQL:

2009-06-29 18:14:58,857 Trac[loader] ERROR: Skipping "trac.db.postgres = trac.db.postgres_backend": (can't import "ImportError: No module named _subprocess")

The functional tests didn't catch this, as for running the functional tests with 2.3 we require to have a backport of subprocess ;-) (see TracDev/FunctionalTests#Dependencies).

[rblank]

We have several options (presented in my order of preference):

1. Bundle the subprocess module with Trac. Would the licenses allow that?
2. Require the subprocess module for 2.3. Not really an option, I guess.
3. Import subprocess in the backup() functions instead of at module level. Users of 2.3 will still have to use --no-backup, as they do now.
4. Rewrite both backup() functions with NaivePopen.

Thoughts?

[mixedpuppy]

I like 3, wrapped with a try/except to output an error message telling the user to install subprocess if it is not available

[cboos]

I was going the 4. route, but as the patch doesn't look pretty, I vote for 3. as well ;-)

[jonas]

I've implemented option 3 in [8338] and [8339] and it seems to work fine.

[rblank]

As there are already two regressions in 0.11.5rc1 (one fixed in [8343], the other one currently discussed in #7823), I assume we are going to do a second release candidate.

I would propose to include the following items related to #7715 in 0.11.5rc2 as well:

• Fix the merge property renderers for scoped repositories (comment:70:ticket:7715)
• Split the merge property renderers into their own component (comment:67:ticket:7715)

The second item allows disabling the renderers in case performance is too low (it is fine here on t.e.o, but eblot reported that it was too slow for hundreds of merge sources), without having to disable the renderers for the other svn: properties.

Opinions?

[osimons]

As for #7823, the last inline patch seems to have fixed it for a user (ie. reverting to the same egg-cache behaviour for the workaround as before [8335]). Apart from that, cmlenz hinted at possibly being able to release Genshi 0.5.2 soon (Tuesday perhaps). If that happens I vote in favour of bumping requirement for 0.11.5rc2 to Genshi 0.5.2 and dropping the workaround code completely.

I'll be mostly offline this week (holiday at the cabin), but should be able to follow progress and discussions + likely be online a bit in evenings CET.

[rblank]

Replying to rblank:

I would propose to include the following items related to #7715 in 0.11.5rc2 as well:

Both patches have been applied to 0.11-stable.

[rblank]

A question to the testers on this ticket: have you found any major issues with 0.11.5rc1, apart from what was fixed in [8343-8354/branches/0.11-stable]? Or is everything working fine for you?

[mixedpuppy]

So far for us, no issues discovered, osx and centos as primary operating systems. -shane

[rblank]

A quick note in case anybody here has missed the announcement: 0.11.5rc2 has been released yesterday. Please update your installations! 0.11.5 final is expected to be released sometime this week.

[Simon <saimen54@…>]

So far no issues.

[rblank]

#8475 has popped up and is due to changes since 0.11.4. At first sight, I'm not sure if it's an issue with Trac or flup. I'll have a look tonight.

[rblank]

#8475 indeed seems to be an issue with Trac, which was hidden by BaseHTTPServer.HTTPServer accepting integers as header values and converting them to string.

The patch is very simple and low risk, so I would suggest including it in 0.11.5 even without going to the trouble of a 0.11.5rc3.

Thoughts?

[osimons]

Patch looks trivial and correct. Don't think it warrants a new RC by itself, and if nothing else appears before Jonas heads over to his release checklist I'm +1 on just committing that patch and releasing 0.11.5.

[Thijs Triemstra <info@…>]

Updating to rc2, rc1 worked fine.

[rblank]

The patch from #8475 was applied to 0.11-stable in [8358].

[jonas]

I agree, [8358] should be safe enough to include without any additional release candidate.

[jonas]

Yay

[cboos]

gg ;-)