[Chandler-dev] Moving forward with the wx on the trunk

Heikki Toivonen heikki at osafoundation.org
Thu Apr 13 22:01:59 PDT 2006


Philippe Bossut wrote:
> That's assuming that John can repro the failure on all platforms
> consistently. This afternoon however John told us that he couldn't repro
> the failure on Windows with tarball 39, only on Mac. How is he going to

Yeah, Windows Tinderboxes were running fine with wx 39. Mac and Linux
are the platforms that were having most of the issues.

> make sure when he identifies and fixes the Mac issue that Windows and
> Linux will pass FT? He'll have to see what happens on Tinderbox. If he
> makes one single big merge at the end and FT fails on Windows (high
> possibility IMHO), we wouldn't have made any progress.
[...]
> If "any commits either on the trunk or branch won't affect that" why
> would incremental updates?

I think we have some confusion of the expected way to proceed.

Here's what I think we agreed on and why the wx39 branch was created in
the first place:

* John tracks down the revisions that caused the problems. There are no
incremental checkins when he is tracking down the issues. The issues get
fixed on the wx 39 *branch*.
* John et al check in their alpha2 fixes on *trunk* so that we can
*test* them ASAP
* Once branch is working, it becomes the *new* trunk. The alpha2 fixes
we need will either happen while it is still a branch or right after it
has become the new trunk (but note that we have already done most of the
testing so there shouldn't be new surprises).

Note that my understanding of the plan is that there will be no
incremental checkins anywhere. John uses the past svn revisions to find
the bad checkins in his local tree. He does *not* redo the incremental
checkins again on the trunk to find out when things break.


I think what you are thinking/proposing is that John redoes the
incremental wx changes on the trunk, checkin things in as he goes.
However, I don't think this is a good idea for a few reasons:

* First of all, if we did this we wouldn't need the wx39 branch at all
* If we did this, it would make it more difficult to test the alpha2
fixes (we'd have to wait until all wx issues are fixed)
* Redoing the old checkins is error prone

-- 
  Heikki Toivonen


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 251 bytes
Desc: OpenPGP digital signature
Url : http://lists.osafoundation.org/pipermail/chandler-dev/attachments/20060413/8e8c2727/signature.pgp


More information about the chandler-dev mailing list