[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: dfa.h / dfa.c diff versus gawk attached

From: Aharon Robbins
Subject: Re: dfa.h / dfa.c diff versus gawk attached
Date: Mon, 22 Oct 2007 22:06:05 +0200

> Date: Sun, 21 Oct 2007 18:08:35 -0500
> From: address@hidden (Karl Berry)
> Subject: Re: dfa.h / dfa.c diff versus gawk attached
> To: address@hidden
> Cc: address@hidden, address@hidden
>     So, sorry. For now, my source for regex is GLIBC.  If / when gnulib
> If that works better for you, fine.  It makes little difference in terms
> of the actual code.
> For anyone using gnulib already (as I hope grep will), regex might as
> well come from there too.

Makes sense.

>     regex can be dropped into gawk without all the goo, I'll reconsider.
> Not going to happen.  Part of the purpose/benefit (obviously you don't
> find it such :) of gnulib is all the autotools glop and goo.  Although
> I'm a little surprised regex was tinkered with to such an extent, but I
> guess I shouldn't be.

The problem I've seen with gnulib is that in order to get one piece, you
often end up pulling in some large transitive closure of other pieces
by the time you're done.  For gawk, I suspect I could end up with all
the support goo severely rivaling the size of the actual program. Not
what I want.

I like the idea of gnulib, I really do.  But the pieces are not
independent enough.  I can, and do, go to GLIBC to get the occasional
small building blocks I need. It's generally easier than using gnulib.
I have to wonder if I'm not the only GNU maintainer who feels that way...

>     In the meantime, gawk's regex also has some fixes vis a vis glibc that
>     I don't think are in gnulib, but I don't have the cycles to do the merge.
> Life goes on.  There have been umpteen versions of regex forever, I
> don't suppose the situation will ever be resolved.

Sigh.  We've discussed this off-list before... A separate regex maintainer
is needed, someone who actually understands all the theory behind DFAs and
NFAs as well as the real-world issues of character sets and encodings. It's
a tough job.  I'm willing to help, I'd like to see one canonical version.
But I can't do the job.  (We need to find some bright, _single_ person, straight
out of CS grad school... :-)

Anyway, this is off topic for bug-grep.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]