Re: reviving sawfish development.

[John Harper <jsh unfactored org>]

On Jun 26, 2007, at 4:07 AM, Janek Kozicki wrote:

> I'm writing to you on behalf of our sawfish mailing list. Based on
> recent discussions about sawfish development and SVN:

(I'm still subscribed to the mailing list, but I rarely read it)

> we have noticed that:
>
> - there are some patches submitted to the list (even some time ago),
>  which did not found their way to the SVN repository.

examples? When I see bug-fix patches that are obviously correct, I try  
to remember to commit them

> - some other are working on fixing bugs (eg with AMD x64 compilation,
>  etc) with no place to commit their fixes.
>
> - every once and then someone comes, says that he plans to improve
>  sawfish somehow, and asks if his improvements will be added to the
>  release.

no offence, but talk is cheap :-)

I think it's better to err on the side of caution here, by its nature  
sawfish is a hard-to-understand project (arcane X details, multiple  
languages, multiple OSes, etc), and so is very easy to introduce bugs  
into. Every non-trivial change has to be very carefully considered.  
Given all that, I consider sawfish "finished" – bugs can be fixed, but  
extra functionality should be added as lisp extensions outside the  
core project. (I also see no need to make "releases" from the  
repository, it's stable enough that grabbing the code from svn is  
equivalent.)

So that's why I think bug-fix mode is better than active development  
for sawfish. But if someone wants to create an active branch in the  
gnome svn repository and work on it, that's fine with me  
(alternatively they could just create their own repository using git- 
svn and publish that directly). If any changes get enough testing  
coverage and look sane, they can be merged back into the trunk,

	John