... and this kind of discussion is precisely why invention by standards
committees is a Bad Idea. It's impossible to evaluate the merits of this (or
any) proposal without an understanding of its implications for application
development; and you can't really do that without a trial implementation.
No matter how good the proposal, and with all due respect and thanks to
Ulrich for fulfilling his assigned action, I believe we should refrain from
adding this interface to the draft standard. Unless and until something has
been implemented, standardization is inappropriate.
Or is there some reason invention is OK in a joint ISO/PASC/Open Group
context but not in a purely POSIX?
Jason
-----Original Message-----
From: Ulrich Drepper [mailto:yyyyyyy@xxxxxxxxxx]
Sent: Monday, August 14, 2000 1:57 PM
To: yyyyyyyyyyyy@xxxxxxxxxxxxx
Cc: yyyyyyyyyyyy@xxxxxxxxxxxxx
Subject: Re: AI 2000-05-010: proposed interface
David Korn <yyy@xxxxxxxxxxxxxxxx> writes:
> If it is true, that no two characters can have the same collation
> order, (which I assume is required), then wouldn't == mean that the
> strings are equal?
Yes.
> Clearly this can be written in terms of yours and vica versa,
> but this seems more likely to be more efficient to implement.
> It could even be done as a table lookup.
Speaking from experience, implementing the string comparison function
in terms of your function is far less efficient. You have to look up
in various tables and possibly ceate data structures. Also, this is
the more consistent interface since there is strcoll().
--
---------------. ,-. 1325 Chesapeake Terrace
Ulrich Drepper \ ,-------------------' \ Sunnyvale, CA 94089 USA
Red Hat `--' drepper at redhat.com `------------------------
|