Re: PostgreSQL 9.3.5 substring(text from pattern for escape) bug

Andrew Gierth <andrew(at)tao11(dot)riddles(dot)org(dot)uk> writes:
> "Tom" == Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
> Tom> Huh, interesting. So we should be translating the initial
> Tom> substring to a non-greedy pattern. I believe Spencer's engine can
> Tom> handle that by sticking (?:...){1,1}? around it.

> Your suggested fix doesn't seem to work. If the leading/trailing
> substrings do not have | or parens in then it seems to work to wrap them
> in (?:(?:)??...), thanks to the rule that the first quantified atom in a
> subexpression sets the whole subexpression's greediness, but handling |
> or parens correctly seems harder.

[ pokes at that... ] Huh. That's a bug, which AFAICS is aboriginal in
Henry's code: it optimizes away a {1,1} quantifier without regard to
whether the quantifier is attempting to impose a different greediness
preference than its argument would have naturally. The attached
seems to fix it.

regards, tom lane