[Qgis-developer] Release plans for QGIS 0.11.0 and on to 1.0

[Martin Dobias]

On Wed, May 28, 2008 at 11:25 PM, Tim Sutton <tim at linfiniti.com> wrote:
>>
>> What about further roadmap?
>
> I think the last time we discussed this, the general concensus was that we
> would put all our focus into 1.0 for now and after 1.0 goes out, have a
> period of planning for 2.0. I would hope that we can limit the features to
> what we can achieve over the course of 1 year or so, with a couple of
> preview releases during that time.
>
>> Schedule of 1.x versions?
>
> During the same period as 2.0 development we should aim to keep our more or
> less 2 month cycle running with bug fix releases to QGIS 1.0. In my opinion
> we should try to identify an official maintainer for 1.0 (like the Linux
> kernel folks do for each release).

I see basically two possibilities:
- continue development of 2.0 in trunk and leaving 1.x in a branch.
Changes made in every round of refactoring/redesign will be merged
back to trunk. It can be expected that devs will be developing new
features mainly against trunk, 1.x branch will get just necessary
fixes, possibly backports of some new functions.
- 1.x versions will be developed in trunk and redesign will happen all
time in branch. After some time, commits from trunk will be
practically unmergeable to development branch without great manual
modifications. The development branch will be probably used and tested
only by very small amount of devs.

I'm much in favour of using the first approach - let's conserve QGIS
1.x in a usable state without bringing many new features in 1.x cycle
and go towards advanced architecture in 2.0. This will probably have
impact that users won't see any new "stable" release for longer time,
but it will be worth the changes.

>> Are we going to
>> maintain stable ABI?
>
> I'm not experienced with the in's and outs of what is required to maintain
> ABI compatibility within a C++ / Qt project. Could you point to any good
> resource on this.

Now I'm unable to point you to any good resource discussing this. But
having ABI stability for C++ libs is quite hard to achieve (see
http://www.murrayc.com/blog/permalink/2007/03/12/abi-stability-of-c-libraries/).
Anyway, we don't even have to think about that. There are just few
plugins or applications in c++ based on QGIS libs, so recompiling them
shouldn't hurt anyone.

>> API?
>
> My feeling is that we should maintain API compatibility for the 1.x
> lifecycle so that poor plugin etc developers dont need to work against a
> moving target any more.

Yup, stable 1.x releases should keep API compatible with 1.0.


>> I'm sorry but it looks like I won't be able to participate - June is
>> full of exams and visitors and then moving back to Slovakia, from
>> mid-July I'm on holidays in India until mid-September. I could
>> participate during the autumn...
>
> :-( I was looking forward to getting some hand-standing tips :-P Well
> hopefully we can arrange the next one so that you can attend, or perhaps
> there is some way to get you out to SA for FOSS4GEO2008, where a number of
> other QGISsers will be attending too.

Don't worry about hand-stands :-)
However there's still a chance that I could attend for 1-2 days - in
window between 24. when I have my last exam and 29. when I'm leaving
from VLC.


> I must say it is really cool to see how much the python work you have done
> has spawned a developer community of its own. I was actually wondering if we
> shouldnt start a qgis-python mailing list since it seems to be an area of
> QGIS with a life of its own...

I'm also quite positively surprised how much buzz PyQGIS has created.
Launching qgis-python list isn't neccessary so far...

Cheers
Martin