2010/4/24 Doug Blank <doug.blank@gmail.com>
On Sat, Apr 24, 2010 at 12:08 AM, Rob Healey <robhealey1@gmail.com> wrote:
> Dear Doug and Stephane:
>
> Since this issue is being linked and directed at me, may I ask what was
> messed up with NarrativeWeb please?  I know that I did a major commit with
> many things being addressed in one commit, but I do not know what I did to
> mess it up!!!

Rob,

I guess that is the main point... no one knows exactly why things were
broken, nor what to do to fix it. There was no way to pinpoint the
change that caused the breakage because it occurred in a monolithic
set of changes. With Narweb, we don't have a test suite to run through
to know all of the things that were broken. Maybe it was just a single
bug, or maybe there were a host of things wrong.

When we are near a release, we usually tread lightly, but still fix
things as they come in or are identified. We know that at any minute,
we might have a release from SVN. At every commit, we have a belief
that the codebase is just a little bit better than before. If we have
any doubt about that (that a change might have an adverse affect that
we might not understand) then we don't make those kinds of changes.

Narweb was in a slightly different state. I didn't realize how many
regressions we had from 3.0. It is understandable that we had so many
issues: the code was rewritten, and is much better. It would be nice
in the issue tracker if one could identify (and filter on) those
tickets regarding a single piece of code, like Narweb. That would have
made it clear that:

* a lot of people rely on Narweb
* there were many things wrong that people need fixed

The fact that we had a 3.1 release and I didn't know that suggests
that we need to make some changes, such as better tracking. You should
feel confident in asking for help if there is an issue that you can't
(or don't want) to do. We need to do a better job in assigning issues
as they come in, or at least being aware that there is a regression
that no one is working on.

What happened was a worst case scenario---we just got really unlucky
that your commit happened right before the release grab. But, we'll
make sure that luck has less to do with it next time:

1. branch is sacred and stable---no change is made unless it fixes something
2. all changes are tested
3. releases will be announced days in advance
4. no release will be made on the heels of any change
5. we'll work together to prevent regressions
6. we'll break up large pieces of code into smaller plugins (plugins
for plugins)
7. we'll have ways (addons) to be used to override core plugins

This was a learning time for all of us. No hard feelings, as long as
we learn from this. Now, back to work :)

To answer your queation Rob.

There where some new bugs (Doug said in image region code) and reintroduced bugs (stephane said you removed one of his bug fixes, this would explain why the copyright message of Stepane -# Copyright (C) 2007-2010  Stephane Charette - was changed to 2007-2009 in your commit). There where bugs in 3.2.0 also, so that is annoying, but 3.2.1 would still have been better than 3.2.0 even with those bugs. Note that just looking at the output of svn diff should have clearly shown you that 2007-2010 was changed in 2007-2009, indicating the patch contained wrong code (perhaps you resolved a conflict on svn up wrongly?).

The big problem however is that you have removed in the commit 

from __future__ import with_statement

breaking nar web on python 2.5. Windows and Mac are still mainly using 2.5, so nar web stopped working for them (they have problems with bsddb being too slow in python 2.6).
This leads me to believe you have just copied a piece from trunk to branch32.
For version 3.3 we will require python 2.6, so windows and mac must make sure they can use python 2.6+ next year when we release 3.3, and so it is ok to remove python 2.5 code in nar web in trunk, but it should be in a commit with only code that is relevant to this, and with a commit message like 'Remove python 2.5 support'. And the commit should obviously not happen on branch32

Benny



-Doug

> Sincerely yours,
> Rob G. Healey
>
>
> 2010/4/23 Doug Blank <doug.blank@gmail.com>
>>
>> 2010/4/23 Stéphane Charette <stephanecharette@gmail.com>:
>> > I would like to send something to the gramps-users list so our users
>> > know what is happening.  But I need help to fill in the blanks.
>> >
>> > ----------
>> >
>> > A last-minute checkin to the 3.2 branch is causing a problem in Gramps
>> > 3.2.1 released this week.  We have pulled this release from
>> > Sourceforge so people don't mistakenly download this version.  Gramps
>> > 3.2.2 will be released shortly to address the issue.
>> >
>> > If you are already using Gramps v3.2.1...
>>
>> ...you might want to upgrade to v3.2.2 next week. If you haven't
>> upgraded yet, you can skip 3.2.1 and go right to 3.2.2.
>>
>> -Doug
>>
>> , ....?  Stop using it?  Don't
>> > use just NarrativeWeb?  Will it crash?  Will it lead to database
>> > issues?  I have no idea...
>> >
>> > The Gramps developers understand the need for a stable release, and we
>> > will continue to refine our development and release process to reduce
>> > the chance of something like this happening again in the future.
>> >
>> > Signed,
>> >
>> > The Gramps Development Team
>> >
>> >
>> > ------------------------------------------------------------------------------
>> > _______________________________________________
>> > Gramps-devel mailing list
>> > Gramps-devel@lists.sourceforge.net
>> > https://lists.sourceforge.net/lists/listinfo/gramps-devel
>> >
>>
>>
>> ------------------------------------------------------------------------------
>> _______________________________________________
>> Gramps-devel mailing list
>> Gramps-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/gramps-devel
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> Gramps-devel mailing list
> Gramps-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/gramps-devel
>
>

------------------------------------------------------------------------------
_______________________________________________
Gramps-devel mailing list
Gramps-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/gramps-devel