Thanks for the prompt response!
Well, I actually had the opposite impression, that the other behaviour is more
common... but for what it's worth, I could not really be sure of this without
checking :-). Anyway, I think that a similar behaviour would indeed be sensible
(and useful).

As for details, except for few languages (which would present exceptions anyway
in different other respects), I am sure that the sets of chars composing the two
groups might be distinguished as follows: 
(1) letters, numbers, underscore, 
(2) all other chars. 
After all, this classically defines identifiers in most languages; even if not
100% of cases are perfectly covered, it would substantially make no difference
in actual usage.
Even the drawbacks should be practically none (except for the work needed to
implement it).
Putting it another way: pleeeeease! ;-)

PS I noticed that the forum font is proportional, which makes my examples
useless as shown... but I see you've got the idea anyway :-)

-- 
<http://forum.pspad.com/read.php?2,32633,32649>
PSPad freeware editor http://www.pspad.com

Odpovedet emailem