[GRASS5] Let's release 5.4 asap

[Markus Neteler]

Hello Paul,

On Thu, Sep 30, 2004 at 03:32:15PM +0100, Paul Kelly wrote:
> Hello Markus
> 
> On Wed, 29 Sep 2004, Markus Neteler wrote:
> 
> >Hi developers,
> >
> >I still didn't read all the messages from the last weeks
> >but would like to suggestion to get out 5.4 as soon as possible
> >(means within days/weeks time frame). There is not much point
> >in not releasing it for two main reasons:
> >
> >- 5.0 is now really outdated
> >
> >- In the community 5.0 is considered to be stable and so it is
> >  heavily used (we saw that in Bangkok). People complained about
> > problems which we already solved month/years ago. On the other
> > hand it is pretty difficult to convince them to change to
> > 5.3 as they say "why is it indicated as development version then?".
> 
> But for a long time now it has said on the website 5.3 is 'testing' and 
> 5.0 is 'stable but outdated' which I feel is a very good discouragement 
> from using it.

I was hoping that as well, but it's unfortunately not true.
Many GRASS users seem to be very conservative concerning version
numbers. I asked several people in Bangkok why they used 5.0.x
and then presented long lists of problems in their conclusions
(most of them are already solved in 5.3/5.7). They answered that
the web pages indicate 5.3 only as testing but 5.0 is indicated
as stable.

> Straight away we could delete 5.0 from the list on the 
> front page leaving only 5.3 and 5.7 I think.

I'll do that if there aren't objections.

> > Software can always be better. Waiting with a 5.4 release doesn't
> > make sense as open things are not going to be solved in the
> > near future (shared libs on Mac etc). So, please, let's get 5.4
> > out and "downgrade" 5.0 in it's importance.
> 
> The thing is that once 5.4 is released we are going to stop CVS access to 
> the grass repository and only work on 5.7. But considering people are 
> going to be going on using 5.3 for many many years we still need to fix 
> everything that was added since 5.0 but is not finished yet.

Oh. This will take another few years :-)

> > To have 3 versions on the web site is considered to be very
> > confusing (I heard that many times the last weeks).
> 
> Well we can take 5.0 off right now I think.

It should also disappear from the download page then.
 

> >Putting much efforts into 5.3 (tcltkgrass etc) might not be worth the 
> >time spent.
> 
> I think it will greatly reduce the volume of annoying complaints from 
> people in the future if we make sure what's there now works, before 5.4.

Mmh. So we better do not look at the bugtracker.
IMHO we need a pragmatic solution. 5.3 is quite good now.
And 5.4 can stay open for bugfixes of course.

At least I would like to get rid of the complicated merge
procedure for 5.7 which is hard to explain to non-experts and
also makes RPM packaging difficult.
For that 5.3 has to be sort of frozen.

> >We have 5.7 and we should concentrate on this version.
> >So, what's holding a 5.4.0 release? Any important problems which
> >will be fixed *soon*? Let me cite these mails:
> >
> >http://grass.itc.it/pipermail/grass5/2004-July/014954.html
> >http://grass.itc.it/pipermail/grass5/2004-July/014955.html
> >"my plan is that when 5.3 can be compiled with shared libraries on all
> > supported platforms, we should make this the default and release it as
> > 5.3.1"
> >
> >We are probably years away from that due to personal time constraints.
> 
> Maybe not as far as it seems. I said to Lorenzo that I may have access to 
> a Mac at my new job now, where I could test out the shared libraries 
> problems on OS X. I think Cygwin is already OK. I just haven't asked the 
> person with the Mac for access yet.

Is this the limiting factor? Other *important* outstanding problems?

> >Instead of holding the 5.4 release for further month/years, we should
> >get it out. Once the shared lib problems are solved we can publish the
> >next subversion.
> 
> I'm not sure about that---I thought there would be virtually no more 
> changes at all to 5.4 after we release 5.4.0. That understanding is what 
> would make me hesitant to agree to a quick release. If that was changed, 
> things would be different.

Again, we need a new version soon which can be called "stable" according
to the version number. After a 5.4 release bugfixing is obviously possible.
But why adding new features to 5.4 if 5.7 is already there? With 10 more
developers it might be feasible to maintain two versions but currently...

Markus