On Tue, 27 Mar 2001, Jarkko Hietaniemi wrote:
> I think the following patch, or its moral equivalent, is needed.
> Saying 'case-insensitive' isn't enough: you need to tell what to
> do if the keys compare equal (in other words, we need two-level
> comparison). This patch makes the test pass in Tru64.
Yes, good one.
Also, should tolower() become toLOWER()? If I'm following the #define
forest correctly, toLOWER() is defined as a macro in handy.h, whereas
tolower() seems to be at the mercy of the underlying C library (according
to my gcc -E).
BTW, I'd have no objections to just stating that the sort order is
indeterminate. After all, I just want it to return a number in scalar
context :-)
Cheers,
-Ben
--
signer: can't create ~/.sig: No space left on device
