Re: GLADE and GPL (was: what DOES the GPL really say?)

[fielding@kiwi.ics.uci.edu (Roy T. Fielding)]

>Well I understand that, particularly in the case of academic people
>fiddling around, it might be quite reasonable to have miscellaneous
>patches floating around, but for our mainstream customers, very
>careful configuration control is extremely important. I really don't
>think any of them would apreciate us releasing a new release of GNAT
>every day! 

Have you talked to your customers about that?  I am certainly aware that
many people prefer stable releases, but making unstable releases available
to those that need/want them doesn't affect those other customers at all.
The only thing you need to do is label them appropriately and make sure
that the labels are understood.

The problem that I run into is that I cannot distribute software that
doesn't compile or run correctly on one of your public releases, even if
I have a private version of the compiler with the corresponding bug fixed.
Also, because the changes between releases are large, it is much more
difficult for me to isolate new bugs introduced between releases.

>By the way, it is quite out of the question to make our problem report
>database public. Many of the problems involve customers proprietary
>code which we protect extremely carefully.

I wouldn't make the entire database public, just those entries that
are public.  In fact, I'd probably use separate databases for the two.
What we need to see is a list of problems (open, in-progress, and closed)
which is kept up-to-date and searchable on-line.  More importantly,
everyone should be encouraged to find and/or fix problems, not just
the contractees, and the only way to do that is to make them public.

>One thing to remember in this discussion is that the continued development
>and improvement of GNAT depends on support from our major customers. As a
>result, it is those customers who determine the requirements, and we find
>that they are much more interested in stablility than in haveing very
>frequent releases.

From my perspective, having founded several community-based software
development projects, I find that statement to be a bit bizarre.  I guess
that is why Eric Raymond calls it ``the "cathedral" model of FSF''.
Continued development and improvement of GNAT doesn't depend on your
major customers -- that can be accomplished without having ACT as the 
high priest.  What your customers need ACT for is validated releases
and guaranteed hand-holding, neither of which benefit from infrequent
private releases vs frequent public releases.

In my opinion, you should be finding ways to let the community help
you support and improve GNAT (what we all want) rather than play one
customer over another as a business strategy.

>It really is hard for me to believe that you think it would be a good
>thing for us to post every patch to GNAT on CLA as soon as we develop it.
>Anyway, we strongly disagree, and are not about to do this under any
>circumstances.

I wouldn't post it to CLA.  I wouldn't even read CLA if there was a
reasonable alternative.  Use the Web.  Remote CVS would be even better,
but that depends on your own development environment.

Anyway, I'll shut up now. I wouldn't have even raised the issue if
you hadn't claimed that ACT support was somehow inherently better than
free software support.  Free software has the capacity to enable an open
development process in which any competent user can become an active
participant, rather than just a passive recipient.  No closed development
group can compete with that.

.....Roy