Re: KDE CVS Commit Policy (3rd version)

On Friday 11 April 2003 08:07 pm, Cornelius Schumacher wrote:

(Arghh, sorry, I was planning on sending this a week ago, but got distracted 
by schoolwork)
> Once again I incorporated the comments and converted the text to HTML
> (see attachement). I will put the page into CVS at
> "developer.kde.org/policies/commitpolicy.html" if there are no serious
> objections.

One objection to a part of point 12; mostly to a specific example which seems 
to imply something different than the section title. I think we should strike 
the  "I don't know what happens, but till somebody finds out, I'm disabling 
it" point. Disabling functionality to avoid serious crashes/bugs 
(particularly bugs with potential security implications) that can't be 
analyzed in a reasonable amount of time for a service release is, IMHO, a 
very sane thing to do; as long as it is fully understood why the disabling 
bypasses the code affected by the problem (even if the problem isn't fully 
understood itself). This should, IMHO extend to pulling applications from 
release when they obviously aren't ready -- as has already happend with KOnCD 
for 3.0. I think this is critical since, being a volunteer development 
effort, we can't throw manpower on a problem -- thus sometimes it may be best 
to cut our losses by slashing a feature..

#19 is violated by at least some files in CVS; the argument in there is also 
made somewhat less compelling by the -kk options to cvs diff. 

Also, I think #17 is enforced by scripts, so it may be useful to add a note 
to that effect. Also, I think the usage of GUI keyword should be made point 
22, since it's pretty important information for the documenters, but many 
people don't know to use it, or exactly when to use it. (I am in the latter 
category - any doc people around to enlighten me?)

Thanks for listening,
Maksim