This is the mail archive of the
mailing list for the GCC project.
Re: [Patch] Support assertions and greedy/ungreedy matching in regex
- From: Paolo Carlini <paolo dot carlini at oracle dot com>
- To: Tim Shen <timshen91 at gmail dot com>
- Cc: libstdc++ <libstdc++ at gcc dot gnu dot org>, gcc-patches at gcc dot gnu dot org
- Date: Thu, 12 Sep 2013 17:04:07 +0200
- Subject: Re: [Patch] Support assertions and greedy/ungreedy matching in regex
- Authentication-results: sourceware.org; auth=none
- References: <CAPrifD=FsT0c6+VK7P6mFO6D9pVUbDdy8Soq-rmHoMA55t7t9A at mail dot gmail dot com> <5231C831 dot 2010206 at oracle dot com> <CAPrifDntDdXcO0Yj9GNZ1BPr_3wwj5-QfVLP=hfXnm8e9MPKEA at mail dot gmail dot com>
On 09/12/2013 04:51 PM, Tim Shen wrote:
Ah, Ok, great. Maybe I simply misunderstood: when you added that
testcase a few days ago, I thought that it simply needed one of the
missing features. In fact it's about a specific "detail", let's call it
like that, one of the simple ones part of the work ahead. Fine with me
On Thu, Sep 12, 2013 at 9:57 AM, Paolo Carlini <firstname.lastname@example.org> wrote:
Great. A quick-quick comment: if these are the last two features, why we
can't un-xfail the testcase which we added latety? Also, a grep revealed a
couple more xfails. Can you clarify?
I say `feature` when I think that, what these xfails reveal are too
small to be features, say, regex_search/regex_match flags. Now turns
out "feature" is not a good word for them. I do mean all C++ library
independent part, or pure regex function part, are done. Next days I
will add flags implementation.
Cool. Really, I remember well that patch going in but I thought it was
only *part* of the work on the other dialects, *not* all of it.
Also, much more generally, I would be curious about the remaining work: I
think essentially it boils down to the vagaries for the other non-default
regex dialects? Is it a lot of work?
already supports all dialects(ECMAScript, basic, extended, grep,
egrep, awk) specified by standard. Most of the differences between
them are eliminated by _Scanner, aka tokenizer, so it's actually not a
lot of work to even add one more syntax.
Eh, let me (us) think a bit about that, we don't want to make legal
mistakes. With boost should be more straightforward, we already borrowed
shared_ptr for example. First, you could send an email to the author of
the boost implementation and raise the issue. Of course add me, the
other v3 maintainers in CC, speak on behalf of the project not just of
yourself as the actual implementor of our code.
But again, I think more
testcases are needed, especially from those experienced regex users.
I'm actually not a big fan of regex(but of NFA ;), or can I borrow
some boost/libc++ testcases without making any licence trouble?
Also, I'm thinking, when those xfails are gone, and the small "details"
correctly handled, we should have a news entry on the GCC mainpage: will
be also useful to attract testers.