Front page | perl.perl5.porters |
Postings from September 2011
On Sun Sep 04 13:20:10 2011, tom christiansen wrote: > "Father Chrysostomos via RT" <perlbug-followup@perl.org> wrote > on Sun, 04 Sep 2011 12:15:50 PDT: > > >How do you defined âunusedâ? How many people know what version of > >Unicode their perl and their editor (actually, their fonts) are using, > >or whether they are even the same? What if I use a new character as a > >delimiter, not realising that perl considers it to be âunusedâ, and the > >next upgrade will break my script? > > I don't understand what you're so worried about. The Unicode > Pattern_Syntax character property has strong stability guarantees. The stability guarantees do not guarantee anything if one’s editor and one’s perl installation have differing Unicode versions. > Why not use just the open/left thingies in BidiMirroring that are > pattern syntax as openers and the corresponding mirrored bit for > closers? Point this at BidiMirroring.txt... My point was that I might not be trying to use paired delimiters at all. If I have shiny new fonts with something that looks like a nice delimiter as the glyph for U+10F001 (some time in the future), but my perl installation has the *previous* version of Unicode (before the one that introduced U+10F001), then a perl upgrade may break my code if it turns out U+10F001 is one of those paired delimiters and I was not aware of it. This is not FUD, either, as there was a CPAN module that had to change to work in 5.14, because of the way Unicode identifiers are parsed. So use of any Unicode (non-ASCII) outside of comments and strings is going to cause problems. I don’t know of an elegant solution to that, but until we have such a solution, I don’t think we should spread the problem further by introducing paired delimiters. Actually, I do have an elegant solution: Provide a plug-in mechanism that allows *modules* to do their own delimiter pairing.Thread Previous | Thread Next