[GRASS5] GRASS development roadmap proposal

[Paul Kelly]

On Thu, 17 Jul 2003, Bernhard Reiter wrote:

> Thanks for the proposal, it is a good idea to work on a roadmap.

Yes I was going to put forward some ideas but I have been busy. But this
is just like what I was thinking about; it is a good plan that takes
everyone's comments into consideration. 

> 
> On Thu, Jul 17, 2003 at 12:33:43PM +0200, Markus Neteler wrote:
> > Users regularly ask (me) about the upcoming GRASS
> > releases and their features. To help them and also us
> > we should define a roadmap which also solves to some
> > extent the versioning problem.
> 
> We need to think in development lines 
> or as I often write series. 
> Your proposal below is not written to show the development lines.
> 
> Certainly we should have 5.0.3 as bug-fix releases.
> We might also have further 5.0.x releases which IMO 
> should also be bug fix releases.

There are 3 development lines; this is one of them.

The next development line is the 5.3.x experimental series, which I think
can just correspond to the 5.0 CVS HEAD: there is no need to make another
release branch as there is not going to be any further development in the
5.0 CVS tree after this release (5.4 is the final release from the 5.0 CVS
tree).

I think this is a really good idea and the first 5.3.0 experimental
release should be very soon (straight from the CVS HEAD). There used to be
experimental releases every month or two before the 5.0.0 release but
there haven't been any at all for about a year now. I don't
think the experimental developments (datum support and new NVIZ etc.) are
getting tested very well the way things are now.

And the third development line will be the 5.7.x experimental series. I
think the gulf in functionality between 5.0.x and this series (currently
known as 5.1.x) is much too wide for the version numbers to be only 0.1
apart as they currently are. Being 0.7 apart much better reflects how
different the two versions are (in fact putting the sites back into 5.7
will bring them closer which is probably good, to encourage more people to
make the jump to the new series).

> 
> 
> > Two problems have to be addresses:
> > - in a 5.0.x version we should not introduce
> >   datum transformation as the results will be
> >   different
> 
> I did not follow the datum discussions in detail.
> What do you mean with: The results will be different?

For the re-projection modules r.proj, v.proj and s.proj, the results will
be different if you are re-projecting between two locations that use
different datums (usually points move by a few hundred metres on the
ground). If you are just projecting between lat/long and a
projected co-ordinate system on the same datum, the new datum
transformation functions will not make a difference to the result.


> 
> > - we should release datum transformation as soon as
> >   possible as it is crucial for GIS data processing.
> 
> Another idea would be to make in an experiemental add-on package
> to 5.0.x if the datum transformation support is ready for this.

Yes it is ready, but the changes are too fundamental and affect too many
different parts of GRASS for this to be a viable option.

> 
> 
> It is clear that the 5.0.x serious has to transform into the
> 5.1.x serious which then transforms into 5.2.x at some point
> where 5.0.x support hopefully can be discontinued.
> 
> The renaming part (5.0 HEAD -> 5.3.0 and 5.1.0 -> 5.7.0) suggested
> will cause confusion. I advise against it.

I think the paragraph above (It is clear ... discontinued) is a
little confusing.
Markus has put forward a viable plan for how this transformation is going
to happen. As I said above, I think the 0.1 difference in version
numbers is too small considering the very big changes involved between the
current 5.0 HEAD and 5.1 development series.

Paul