[GRASS5] Re: GRASS 5.3 release

[Markus Neteler]

On Sat, May 08, 2004 at 09:43:48AM -0700, Michael Barton wrote:
> If you mean the scripts that are in the normal GRASS distribution, a 
> week ago, I would have said yes, this is a problem. However, I was 
> wrong about this. AFAICT, all scripts in 5.7 should work with 
> 5.3--except those that call special 5.7 vector features. 

Right.

> Currently, 
> these are very few. The only thing that needs to be changed is how some 
> (a miniority) are referenced in the tcltkgrass GUI and the man pages 
> for the same scripts. Markus suggested that it would help 5.7 
> development it all essential scripts shared by 5.3 and 5.7 be located 
> in the 5.7 CVS branch and represented by links in the 5.3 branch (I 
> hope I have the terminology +/- correct here).

No, I suggested the opposite (maybe I was not clear in that
personal mail). With one limitation:
But I also suggested to stop 5.3 development and to go for 5.7.

> I could  try to get some 
> idea of how many scripts need tcltkgrass updates to use the 5.7 version 
> over the next week if that is helpful.
> 
> For other scripts--i.e., those that individual GRASS users have 
> developed but are not part of the regular release--some that may work 
> in 5.3 may not work in 5.7 without some work but usually not a lot of 
> work. This is not an issue for the release. IMHO, it would be a nice 
> idea to have a script-writing tips page on the GRASS www site to lay 
> out a set of basic GRASS script standards--e.g., bash scripting, use of 
> g.parser headers, etc.--to get all scripts to a 5.7/5.3 compatible 
> format. Perhaps I could try my hand at that this summer, along with the 
> other related things.

Yes, a sort of automated script repository in the web might be quite
useful.
 
> Finally, there are some 5.3 scripts that don't yet work in 5.7 because 
> they call GRASS modules that don't (yet) exist in 5.7. A good example 
> is r.fill.nulls. It requires s.surf.rst. Because of the demise of the 
> GRASS sites format, s.surf.rst does not exist in 5.7.

It's in fact much simpler:
Once we have fixed the RST LatLong bug (the next days, already
identified), I'll submit r.fill.nulls in 5.7 with the small change
that s.surf.rst is named v.surf.rst in 5.7.

> And, unless 
> something has been done in the last couple weeks while I've been 
> focused on the end of semester chaos, there has yet to be a program 
> using the vector-format points to take its place. Somehow, there needs 
> to be a way to sort through these and say which are worth updating to 
> 5.7 and which are not. Again, however, this could be treated as a 'bug 
> fix' later and shouldn't have any effect on the 5.3 release.
> 
> Michael

 Markus