> I don't really like hashing pointers -- they're too regular -- but if
> we need to, I suppose we can.

I do understand why noone *likes* to bump libc's major. I don't see why it
would be bad to rename the new, thread-safe version to libcmt or something
similar (which exploits my intimacy with NT, I think).

I just don't understand how anyone could even consider wasting performance
by doing a thread safe stdio via hashes to file handles. It's probably not
much performance, and if you are lucky you would block on the found lock
anyway, so it doesen't realy matter in the lock case. Isn't the non-blocking
case the common path?

This IMHO is a hack (like the famous aha1542 specific bounce buffer support,
that was kept out of NetBSD and delayed usability of this adapters until the
right-thing-to-do was implemented about two years later, and those devices
had disapeared from the market).


Martin