On Sunday, December 29, 2002 Bram Moolenaar wrote:
> Giuseppe Bilotta wrote:
>> Is it too much to ask a regexp based iskeyword, as an (extra) option?
>> (iskeywordregex) This would be very important for languages where
>> some keyword identifiers can only appear in certain places of a
>> keyword: this is for example the case for TeX's \ or Texinfo @;
>> also, in some natural language, a trailing ' is part of the
>> previous word, for example.
> Currently 'iskeyword' works for too many situations at the same time.
> It is indeed desirable to have a separate definition of keywords for
> different commands. For example, for Insert mode completion and
> recognizing words in the text for commands like CTRL-] and "*".
> This is in the todo list, but it doesn't have a very high priority.
This might be in the todo list but is not what I was thinking
about :) It's an interesting idea, though.
> A rule-based mechanism (to include a character when it is in a certain
> context) would be very complicated.
Even if done by using regexps? For example, in TeX a keyword could
be defined as
The point was shifting from keyword chars to keywords altogether
(at least for keyword-oriented movements).
Giuseppe "Oblomov" Bilotta