Re: Make \< and \> match symbol boundaries

[Stefan Monnier]

> Many characters have symbol syntax in text mode, including -.
> For this reason, the change you have requested would surely
> cause user confusion in that mode.

I don't think it would cause much confusion and we can simply
tell them to use \b in place of \< or \>.
The current situation not only causes confusion but makes it
surprisingly difficult to search for occurrences of the function
`foo' without also matching `bla_foo_bar'.

> I think it would be better to introduce some other way of
> specifying a symbol boundary search.  Perhaps \[ amd \].

As Paul Eggert pointed out, this is already used.  Also,
it's hard enough for people to remember the few existing
constructs without adding yet more.

Paul also said:
> Here's another idea: generalize the existing regexps \b, \B, etc. to
> support arbitrary classes and syntaxes, not just word syntax.  For
> example, to generalize \b we could add the following regexps:

That's an idea.  But note that what my new \< would match is
not "an empty string after a non-symbol char and before a symbol char"
but "an empty string after a `non-symbol, non-word' char and before
a `symbol or word' char".

So it would have to be specified with something like \-s^w_-w_-
(assuming the syntax used is \-s<before>-<after>- and the <before>
and <after> can use ^ for negation).

Things along those lines have already been discussed (there was
also a proposal to allow [:syntax:<...>:] as character classes),
but it's still not clear to me which is most useful, whereas
I know that changing \< and \> to match symbol boundaries rather than
word boundaries is a very simple change and it would be useful
in many situations.


        Stefan