Re: strange highlightings

[Vladimir Volovich <vvv@vvv.vsu.ru>]

"KG" == Kai Großjohann writes:

 >> well, the URL was guessed obviously wrong: the correct one is
 >> 10.95.3.38 but not 95.3.38. And often the text which has nothing
 >> close to URLs is highlighted and buttonized which seems like an
 >> overkill (and a bit annoying :). is there a config option to
 >> restore the previous behavior? 8-)

 KG> The question is, how to recognize text which is a URL and how to
 KG> recognize text which _isn't_ a URL?

 KG> Gnus currently looks `foo.foo.foo' type strings, with three words
 KG> and two dots.  And obviously, digits are allowed in the words
 KG> (www2.frob.org might be a valid URL).  Well...

Khm.. URLs can well be of the form foo.foo.foo.foo and even with more
words separated by dots.

Well, i really don't like this behavior. Why should one assume that a
given string which looks like foo.foo.foo is an URL and moreover, why
should it be assumes as an URL for HTTP? (why not ftp? why not telnet?
why not ICMP?)

It looks like in a lot of cases this new empirical behavior will give
incorrect results (i.e. which will misleadingly interpret some random
strings as HTTP URLs -- e.g. section numbers like 1.2.3, 1.2.4, etc).

 KG> Have you got a suggestion on improving the rules used by Gnus?

I'd rather prefer to reduce this empirical algorithm to previous form
(when only strings starting with ftp:// https:// http:// etc were
buttonized). At least, it would be nice to have an option to restore
that original behavior, e.g.

(setq gnus-buttonize-naked-urls nil)

(which would extend gnus-button-url-regexp if non-nil; i'd prefer to
have by default old behavior though)

Best,
v.