Re: [Proposal] Regex documentation

2001-11-26 Thread David M. Lloyd
On Sat, 24 Nov 2001, Bryan C.Warnock wrote: > On a side note, do we want to be positive-oriented or > negative-oriented? (Do we want to succeed as fast as possible, or fail > as fast as possible? I'm thinking fail, since you're probably more > likely to do that.) If you weight each kind of rege

RE: Re: [Proposal] Regex documentation

2001-11-25 Thread Brent Dax
Dan Sugalski: # Sent: Sunday, November 25, 2001 11:48 # To: Brent Dax # Cc: Angel Faus; [EMAIL PROTECTED] # Subject: RE: Re: [Proposal] Regex documentation # # # On Sun, 25 Nov 2001, Brent Dax wrote: # # > Unfortunately, because Parrot currently has no GC system # and thus leaks # > all ov

RE: Re: [Proposal] Regex documentation

2001-11-25 Thread Dan Sugalski
On Sun, 25 Nov 2001, Brent Dax wrote: > Unfortunately, because Parrot currently has no GC system and thus leaks > all over the place I'll see about fixing that. > and for some reason trying to mem_sys_free() the RE > handle's stack sometimes segfaults the interpreter, I know what that problem

Re: [Proposal] Regex documentation

2001-11-25 Thread Angel Faus
> Okay, I've got another argument for you: speed. > That was a killer one. > > C:\Brent\VISUAL~1\PERL6~1\parrot\parrot>rebench.pl > Explicit backtracking: 75 iterations, about 30.5380001068115 > seconds. > Implicit backtracking: 75 iterations, about 655.510995864868 > seconds. > > The nu

RE: Re: [Proposal] Regex documentation

2001-11-25 Thread Brent Dax
Okay, I've got another argument for you: speed. Basically, I took my current RE code, copied-and-pasted it, renamed the ops a bit (s/rx_/rx2_/g), and modified each op to reflect the new backtracking and push/pop meanings, doing the minimum amount of work on each op that I could get away with. N

Re: [Proposal] Regex documentation

2001-11-24 Thread Bryan C . Warnock
On Saturday 24 November 2001 05:47 am, Brent Dax wrote: > Bryan C. Warnock: > # But given that, you need only define and forward and reverse > # implementations of each op, and that will cover *all* regex > # operations. > # Forward, reverse, centrally-oriented, look-ahead and > # look-behind - t

Re: [Proposal] Regex documentation

2001-11-24 Thread Bryan C . Warnock
On Friday 23 November 2001 10:10 pm, Angel Faus wrote: > =item resubst s, s, ic > > Works like I, but it also declares that we want to do substitution > and that the replacement string is $2. By rematch's description, wouldn't that be $3? ($1 is the string to match against, $2 is the compiled