[GRASS5] GRASS development model Was: A private conversation

[Bernhard Reiter]

Hello everybody,

as you probably know I am with Intevation GmbH 
the company providing the CVS Server for GRASS development. 
Since about 18 month ago we help Markus to organise 
the GRASS development process better. This is a continuing success story.

GRASS as it presented itself to me in mid 1999 was a mess, not even
having clarified license terms. The release and development process
was almost stalled due to Markus being completly overloaded with
GRASS code maintance work and the low number of contributors.

Since then we have made nice progress towards a stable GRASS 5
release and a stable development model. Much code cleanup was done.
The license clarified to be free software, non-compliant code
removed. The development process is open and attracted qualified
developers from all over the world. Release frequencies of beta code
has gone up, so has the overal stability and code quality.
All this is unthinkable without the free software development
process we have now. And the best is, it is also a lot more fun for
everybody to participate.

I am happy to have these discussions about the development model,
because we are open to any suggestion and I think most developers
are proud about how it is working right now.

Of course there is a room for improvement:

	We need a more formalised procedure for testing GRASS
	releases. (And we need more testers!)
	Without testers there is no chance for a fast release
	of a version which can be called stable.

	We would need another release manager for the stable
	code branch, which assists Markus in verifing that only
	critical bugs are fixed and go in the tarball.

	We need more alpha testers.

Of course I agree with most of Markus comments in his last mail.
Anybody who thinks he can help us with GRASS development or can do
better is free to try, so only constructive critism is helpful.

On Mon, Feb 19, 2001 at 12:20:16PM +0100, Andrea Aime wrote:
> Mr Sherpard looks at GRASS from a strict user view, it uses GRASS as
> a product and of course he wants a stable one to make its daily work.

> My only advice for an engineering point of view is: release often, 
> release stable code, release few changes at a time. 

True, but CVS does that for you already.
We just need somebody to actually recheck all changes.
Most developers are willing to work hard to fix bugs, but sometimes
they still introduce another, fixing three others.

> From a user point of view, GRASS 5 needed too much time to get
> released, and this is because it introduces too much changes from
> GRASS 4 -> nviz, fp raster, new sites format, new xdriver
> approach, better portability, etc, etc. 

True.
Right now Markus cannot handle the load to maintain a development
and stable version of GRASS 5. If we do not keep developers 
motivated we will miss very important bug fixes and additional
features. 

	Bernhard
-- 
Professional Service around Free Software                (intevation.net)  
The FreeGIS Project                                         (freegis.org)
Association for a Free Informational Infrastructure            (ffii.org)
FSF Europe                                            	  (fsfeurope.org)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 248 bytes
Desc: not available
Url : http://lists.osgeo.org/pipermail/grass-dev/attachments/20010219/32c67bff/attachment.bin