Georg Baum a écrit :
Am Donnerstag, 30. März 2006 10:50 schrieb Abdelrazak Younes:
I wouldn't like you to be frustrated by people not
having enough time to review your (big and frequent) patches.
This problem would not be solved by rw access. rw access is good for small
obvious stuff that can go in after announcing it, but review is needed
for more complex stuff, and after a consensus on a patch is reached, but
it does not replace review.
I can understand your frustation very well, but I agree with Jean-Marc and
Lars that review is needed. If you want to get a feeling what happens
without review look at the postscript bounding box parsing code in
readBB_from_PSFile() and then try to replace it by something that works.
This is very difficult, because that function is called often and it is
not clear what exactly the callers need.
I am sure that fixing this now is more effort than it would have been to
fix this before it went in, not counting the bug hunting time of many
people who where bitten by this bug.
There is a big misunderstanding here... I have never said that review is
not necessary, quite the contrary actually. I am just saying that we
should find a way to ease the review process. I understand your fear and
I am maybe going to far with what I am proposing. So, another idea would
be to have a "next" branch next to "trunk" where experimental things
could go more easily. In this branch, the proposed patches could evolved
or be reverted. In due time, when the main developers have enough time,
they can review the changes and merge into trunk what they feel are
correct feature and implementation.
For me at least it would be easier (and quicker) to review some change I
am interested in in "next" by reading the SVN logs. IMHO this would be a
very educational way of working because everything would be explained in
the SVN logs, just a few mouse click away (or a few svn command if
you're not a point&click guy :-)).
Abdel.
