Re: Patching patches or rewriting them

[Charles Levert]

* On Tuesday 2005-06-21 at 17:37:33 +0100, Tim Waugh wrote:
> On Tue, Jun 21, 2005 at 12:07:42PM -0300, Tony Abou-Assaleh wrote:
> 
> > If a patch is applied to the CVS, and I find a better solution, should I
> > be writing a replacement patch or a new patch to the most recent CVS?
> > 
> > (For instance, a (supposedly) better patch for the combination of
> > matching-only and context lines.)
> 
> Against the most recent CVS IMHO -- otherwise it's not clear how to
> read the patch, or how to apply it.

I committed a lot of stuff, exactly so that we
can have a common reference for new or existing
developments from now on.  Otherwise, we will
just spend time rewriting patches over and over
again, just to keep up with the many versions
of the other patches floating around, and the
backlog will just keep on growing.

Continuously rewriting patches against several
moving targets can be as frustrating as being
given the same appointment at the physician as 50
other people, but not being told so in advance.    :-)

Let me know if I broke anything, but I have
used a lot of this new stuff for a good while,
and I have verified that "make check" output
remained the same at each step (except when it
fixed things).  I have added a few more tests
in there as well.

Let me know if you would have separated or
combined the commits differently as well, based
on your experience.

Tony:  Have you signed a copyright assignment
       for GNU grep with the FSF?  If you have,
       please submit all the patches you want.
       If you haven't or can't, it may be better
       to submit ideas rather than patches,
       because otherwise the patches would have
       to be rewritten anyway, unless they're
       trivial.  The worst case scenario is that
       you would submit _the_ optimal patch and
       that we would be stuck not being able to
       use it because of this, and thus have to
       settle for something lesser.