On Wed, 24 Feb 1999, Bruce Bowler wrote:
> At 05:16 PM 2/24/99 +0000, Philip Hazel wrote:
> >On Wed, 24 Feb 1999, Bruce Bowler wrote:
> >Aarrgghh. Traditional Unix user names have always been lower case.
> Tradition, bah humbug. He was "non-traditional" to say the least. What I
Imho there's a difference between s/th that's "tradition" and s/th
that's currently the same way it's been for countless years just because
it's reasonable and proper.
The latter seems to be the case with the case-insensitiveness of
<local-part>s, doesn't it?
> how to implement it but what Chad says makes sense as a start as does,
> perhaps, an option to either use getpwent or a "user (or PH) supplied"
> serial read thru passwd routine that does a "truly caseless match" (with
> the caveat that it'll slow things down, perhaps even name the option
> "slow_caseless_match".
Well, generally I had no problem with what Chad suggested (check for
matching lowercase, then check for matching exact-case). But imagine a
mail system processing mails for a very high number of users. Checking for
both lowercase *and* exact case would double the effort - what most likely
would be absolutely unnecessary since probably >90% of mail addresses in
the known universe are lowercase.
>From my experience, a lot of people write to: P.H.Boss@???, knowing
exactly that the correct mail address is lowercase or an alias for
phboss@???. Since they expect the mail system to process the address
in a case-insensitive manner, they choose the more politer option with
upper case initials.
I bet a family-seized pizza that there are no more than 5 admins on this
list generally treating their local-parts case sensitive ;)
Regards,
Volker
--
------------------------------------------------------------------------
Volker T. Mueller Albert-Ludwigs-Universitaet Freiburg im Breisgau
Student der Informatik vtmue@??? +49 761 355-03 -80(fax)
ech`echo "xiun" | tr nu oc | sed 'sx\([sx]\)\([xoi]\)xo un\2\1 is xg'`ol
