Front page | perl.perl5.porters |
Postings from August 2010
[Note: the following reply is based on the consensus reached in previous discussion.] * Abigail <abigail@abigail.be> [2010-08-19 16:40]: > After reading the entire thread, I can't say I really see the > benefits of the proposed (?~) construct. Sure, with enough > imagination, one can think of obscure contructs that will > break. OTOH, in the past, I've written code that actually scans > for (?xism:) constructs and modifies them. Such code would > break if (?xism:) is going to be replaced with (?~:). Such code will break anyway, because the default will soon change to include a new flag that specifies the Unicode behaviour. > I also don't really like the idea of "just enable whatever is > the default on the current version of Perl". '(?si-xm:)' tells > me exactly what's enabled, and what's disabled, regardless of > the Perl version. (?~) doesn't tell me that. It does tell you that: the default is forever going to have the same semantics. However, the flags that constitute that particular match semantic may change over time. (Currently we’re looking at a change Unicode behaviour.) So a (?^:) regex in Perl 5.37 will match exactly the same things as a corresponding (?xism:) regex matched in 5.5, even if the default might be spelled (?xismASDFJKL) in 5.37. So you can write code that looks for `(?^:` and be sure it will never break. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>Thread Previous | Thread Next