[Top][All Lists]

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

Re: [bug-grep] Plan for grep

From: Stepan Kasal
Subject: Re: [bug-grep] Plan for grep
Date: Tue, 8 Mar 2005 10:02:06 +0100
User-agent: Mutt/1.4.1i


On Tue, Mar 08, 2005 at 04:27:28AM -0400, Tony Abou-Assaleh wrote:
> On the long term, I'd like to make -P not experimental [...]

sure.  We will do our best in the 2.6 branch.

I've discussed grep -P with Claudio in private (I didn't want to
disctract the group now).  We came to a conclusion that a patch
to pcre is needed, and I hope that Claudio will help us with this
before we get to 2.6.

Claudio mailed me a very nice description of the problem, citing
some tables from pcre manpages.  It helped me much to understand
the situation.

Claudio, would you be so kind to post here a short summary of the
results of our discussion?
(Feel free to cite from my mails, of course.)

But pcre is *very* powerful, and there can be many problems.
Supposing that we'll get right the issue we discussed with Claudio,
we will have to test heavily the look-forward and look-behind
features, whether they work as expected, even if we are close to
a buffer boundary. As grep is line-oriented, people shouldn't be
able to cross any '\n' with these.
The "option change" feature, eg.  (?m) and (?s), is another scary
thing.  It means that semantics can be changed for _part_ of the
expression.  We should catch these and forbid them.

So it's a really long way, and we shouldn't start it now.  But if
you wanted to help Claudio to prepare for that day, that would be

Have a nice day,
        Stepan Kasal

reply via email to

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