On Mon, Apr 21, 2025 at 12:08:53AM +0000, David wrote: > On Sun, 20 Apr 2025 at 17:42, <tomas@tuxteam.de> wrote: > > On Sun, Apr 20, 2025 at 05:58:31PM +0100, debian-user@howorth.org.uk wrote: > > > > Err, did you notice the bit in that reference that says: "It documents > > > regular expressions in the form available within KatePart, which is not > > > compatible with the regular expressions of perl"? Note that PCRE stands > > > for Perl Compatible Regular Expressions. > > > And note PCRE is not Perl's regexps, but just "inspired by". They > > converged and diverged over their respective histories. To get an > > idea of the current situation, perhaps [2] is relevant. > > [...] > > > [2] https://en.wikipedia.org/wiki/PCRE#Differences_from_Perl > > Thanks for pointing that out. I was guilty of the same misconception > when I erroneously wrote here yesterday: > > > According to [?] it might be [?], which at a glance looks like it uses > > PCRE (ie PERL, contradicting the above statement). > > But there's no contradiction. My mistake was to assume that > PCRE and PERL regex are the same codebase, but that's not correct. > So it's not surprising that they might have some differences. > Thanks for the useful wikipedia page. I know, it's confusing. Perl was a step forward in REs because it left the purity in favour of usability and convenience. It was a step backward because it re-introduced backtracking (which was, since Thompson, a solved problem), thus exhibiting exponential run time in some cases. If you like to disappear in rabbit holes, this [3] one is nice. If not, still the pic at the beginning still shows impressively an exponential runaway for Perl's re engine compared to Tcl's which uses a Thompson algorithm. Note that this was back in 2007 and quite possibly Perl's engine has been refined (it definitely has been extended) since then. Cheers [3] https://swtch.com/~rsc/regexp/regexp1.html -- tomás
Attachment:
signature.asc
Description: PGP signature