Re: "\d{0,4}" doesn't work?

[Michael McCallum <mi...@spinsoftware.com>]

I was doing some work on regexp a while back but it seemed like any changes I might want to 
do would end up making it like oro. So would the work not be better put into Oro?

I like Daniels idea of the posix compliance.
What do you see as the main tasks necessary to fold regexp in as the posix package?

May I suggest a plan ( and yes Jon that means im volunteering to do it or at least some of it ;-)

1) Fix all the current bugs.
2) Make a 1.3 release.
3) Redesign into oro.posix
( Any one have a Posix.2 Standard :). Could just cheat and use rx or something)

Anyone who has submitted a bug (or intends to) could they make sure to provide suceeding and 
failing cases. I can just add those to the test sweet them.

Michael

On 11 Jul 2001, at 9:42, Nino Walker wrote:

> Is there any reason I'd want to stick with this Regexp package?
> 
> After all this discussion, it seems the take away is: DON'T use
> org.apache.regexp.  And, after playing with the ORO applet, it seems much more
> robust.  I'm almost convinced, but ya'll might have some other experiences...
> 
> Thanks,
> 
> Nino
> 
> 
> "Daniel F. Savarese" wrote:
> 
> > Yowser, I didn't realize the project didn't have dedicated workers,
> > but was beginning to suspect.
> >
> > I'm going to have another spurt of time to dedicate to jakarta soon, so
> > I'll try to present a proposal to turn the regexp package into
> > org.apache.oro.text.posix, making its primary goal to provide
> > POSIX-compatible regular expressions.  Don't count on it this time around
> > though because I still need to finish things I promised to do during my
> > last spurt of available time.  Also, someone is going to have to step up
> > to the plate to maintain org.apache.oro.text.posix if I do the initial
> > redesign and conversion.
> >
> > As an aside, it is possible to support multiple regular expression syntaxes
> > through a single matching engine, as is done in GNU regexp, but by nature
> > such implementations tend to have poor performance characteristics.  The
> > jakarta-oro approach is to provide generic interfaces and multiple
> > engines that are tuned to the different syntaxes.  It also provides
> > a foundation layer for building higher level tools.  So an org.apache.regexp
> > compatibility package would be easy to build around the jakarta-oro structure,
> > or for that matter, the JDK 1.4 regular expression implementation, but not
> > necessarily vice versa.  But one of the upcoming issues for jakarta-oro is
> > what higher level tools to build.  For example, the interface presented by
> > the org.apache.regexp.RE class can be made generic across syntaxes
> > by adding a syntax selector, but does it make sense to stick with that
> > interface or is it better to conceive of an equally programmer-friendly
> > but perhaps more resource-efficient approach?  Clearly people like all
> > in one classes like org.apache.regexp.RE, gnu.regexp.RE, and
> > org.apache.oro.text.perl.Perl5Util, but what's the best abstraction to
> > meet the requirements of ease of use, performance, and resource conservation?
> > Anyway, I have design views about this stuff, but am not going to do much
> > about it until we get org.apache.oro.text.regex compliant with Perl 5.6.
> > Everyone is welcome to take the lead on these issues or let me continue
> > to be a bottleneck.
> >
> > daniel
> 



If the brain were so simple we could understand it, we would be so simple we couldn't
-- David Zindell 

Re: "\d{0,4}" doesn't work?

[Jon Stevens <jo...@latchkey.com>]

on 7/11/01 2:12 PM, "Michael McCallum" <mi...@spinsoftware.com> wrote:

> 1) Fix all the current bugs.
> 2) Make a 1.3 release.
> 3) Redesign into oro.posix
> ( Any one have a Posix.2 Standard :). Could just cheat and use rx or
> something)

+1

-jon

Re: "\d{0,4}" doesn't work?

[Jon Stevens <jo...@latchkey.com>]

on 7/11/01 2:12 PM, "Michael McCallum" <mi...@spinsoftware.com> wrote:

> 1) Fix all the current bugs.
> 2) Make a 1.3 release.
> 3) Redesign into oro.posix
> ( Any one have a Posix.2 Standard :). Could just cheat and use rx or
> something)

+1

-jon