[VETO] old ws -- packages, define-behavior

Michael Sperber sperber at informatik.uni-tuebingen.de
Fri Nov 12 08:16:12 EST 2004


>>>>> "Stephen" == Stephen J Turnbull <stephen at xemacs.org> writes:

Stephen> Huh?  Works fine for 21.4 and for Carbon.  What CVS branches can't
Stephen> deal with is if patches are flowing back and forth between their
Stephen> branch and the mainline.  

Yes, and that's exactly what's needed in long-term projects such as these.

Stephen> On the other hand, I don't think that anybody has really acknowledged
Stephen> the fact that Ben's quantity means policy (arguably) should be
Stephen> qualitatively different for him.  For example, you say his ChangeLogs
Stephen> don't allow you to figure out which part of a patch causes a problem
Stephen> because you get a list of 100 file/functions which say something like
Stephen> "Updated to use behaviors".  OK, I can understand that ... but
Stephen> concretely, what do you expect?

The behaviors patch is fine.  A prime example would be the
src/ChangeLog entry from 2002-05-05.

Stephen> Do you really think it's feasible to maintain the 1:1 to 100:1
Stephen> diff-to-log ratio characteristic of "little fix" patches in a 70,000
Stephen> line patch? 

I actually do.  Other projects do it all the time.  Look at the
FreeBSD logs for a good example.

Stephen> On the contrary, it's easy to suggest that maybe a change of that
Stephen> nature should be documented in different form.  Eg, when Ben changes
Stephen> the C code in a major way, he usually also updates the header
Stephen> comments.  He always fixes comments in the neighborhood of his
Stephen> changes, etc.

That's exactly what I have a problem with.  Everyone else carefully
groups their changes to really be related.  This "I was in the
neighborhood, and I saw this piece of garbage" strategy is just awful
for me when I try to figure things out a posteriori.  All these
pervasive changes are usually impossible to roll back after a couple
of weeks.

I'm really hurting here---I would like to contribute meaningfully to
the project, but the way it's going currently, I'm always 2 or 3 steps
behind even attempting to do something.

-- 
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla




More information about the XEmacs-Beta mailing list