Front page | perl.perl5.porters |
Postings from May 2016
On 25 May 2016 at 02:02, Zefram <zefram@fysh.org> wrote: > demerphq wrote: >>I can understand forbidding quantifiers on an assertion, but not >>allowing it if its wrapped. > > My point is that any forbidding of quantifiers on assertions should be a > purely syntactic effect. A quantifier is syntactically permitted on (?:), > and must be allowed regardless of what is contained in the group. If, > instead, the forbidding of quantifiers were a semantic effect, we'd get > into much woollier territory trying to define what semantically counts as > an assertion. (/^/ is, and you'd say that /(?:^)/ is; what about /^(?:)/ > or /^^/?) It'd be much more difficult to define what's permitted in a way > that's independent of the implementation (particularly of optimisation). > Since quantifying an assertion is actually semantically meaningful, > albeit a bit insipid, we'd also run into trouble with machine-generated > or translated code. Keeping the restriction purely syntactic avoids > all of these problems. My first reaction to this is "Ok, I get it, and oh BTW, bonus points for using 'insipid' in a technical discussion". My second reaction to this is that without a quantifier or alternation (?:) is semantically invisible, and should that make a difference? Eg: $ perl -Mre=debug -wle'qr/(?:a)bcd/' Compiling REx "(?:a)bcd" Final program: 1: EXACT <abcd> (5) 5: END (0) anchored "abcd" at 0 (checking anchored isall) minlen 4 Freeing REx: "(?:a)bcd" So I would expect /^*/ and /(?:^)*/ to produce the same compiled optree, and the same errors. But what you are saying is you would expect them to differ. Which I guess is the difference between erroring at parse time, versus optimization time. But the difference makes me wonder a bit, because of things like this: $ perl -Mre=debug -wle'qr/(?:a) */x' Compiling REx "(?:a) *" Final program: 1: STAR (4) 2: EXACT <a> (0) 4: END (0) minlen 0 Freeing REx: "(?:a) *" IOW, under /x we allow comments and whitespace to be inserted between a construct and the quantifier. (I have argued in the past this is not a good thing, but I lost the argument.). So if we determine this at compile time I guess we have to do it *after* we strip whitespace, but before we construct the optree. Historically I have done everything I could to get us out of special casing things based on the spelling of the pattern[1], and to instead use the optree representation instead, so doing the opposite here feels wrong. Although I am open minded about it. Yves [1] I hope this is the same as what you mean by "syntactically". -- perl -Mre=debug -e "/just|another|perl|hacker/"Thread Previous | Thread Next