Reply to message «Re: How to update perl.vim for UTF-8 chars in identifiers? (RFE?)», sent 18:27:40 15 May 2011, Sunday by Bram Moolenaar:
> There has be quite a lot of work on the new regexp code, but the tests > still don't cover enough to be sufficiently sure it won't break > anything. But where is the code itself? > It would be possible to support multi-byte characters for 'isident' and > 'iskeyword' (currently all multi-byte characters are included), then \i > and \k can be used. But: these are global options, we would need a > buffer-local or window-local option for syntax HL. 'iskeyword' is local. But I doubt that this makes sense as we still need to obtain somewhere what characters are included in particular unicode character class, at least without character classes in 'isk' support. Original message: > ZyX wrote: > > I heard something about that vim regexp code needs rewrite/is being > > rewritten (Bram requisted more tests for regexes for new regex engine, > > but I do not know where it is and whether someone is working on it or > > it is just plans). > > There has be quite a lot of work on the new regexp code, but the tests > still don't cover enough to be sufficiently sure it won't break > anything. Also, there has not been an overview of tasks and how much > faster they get. It might even become slower for some. > > > In current engine unicode character classes are not > > implemented, and collections have some weird implementation that will > > prevent you from writing a long collection: according to help [0-9] is > > slower then \d and you can't use [\uN1-\uN2] if N2-N1>255 (so, I > > guess, regex compiler generates code that checks for each symbol > > instead of checking whether symbol is in given range). > > > > So, no, it is not possible without patching vim. > > It would be possible to support multi-byte characters for 'isident' and > 'iskeyword' (currently all multi-byte characters are included), then \i > and \k can be used. But: these are global options, we would need a > buffer-local or window-local option for syntax HL. > > Perhaps using \k gets you close to what you need.
signature.asc
Description: This is a digitally signed message part.
