I thought you were wrong, but now I think you are right (at least wrt Digital
Unix). Here is the Digital Unix man page:
http://alisa.ucsd.edu/cgi/webman?SEARCH+man2+write.2.gz
And the Linux manpages say otherwise (both the traditional writev
manpage, and the new netman manpages for send and friends).
Even if Linux's behavior is wrong, and I personally think our behavior
is a lot better, it now comes down to changing an existing and long
standing Linux API. And just due to that, I am very unlikely to make
the change. There are FTP servers which depend upon this behavior for
single iovec sends, ie. they do check for partial writes and handle
things correctly.
So what if I change the behavior, and someone now writes an app which
now depends on this new writev behavior, now their application is by
definition broken on older kernels and more importantly 2.0.x kernels.
No, I'm not going to advocate nor make this change, sorry.
Later,
David S. Miller
davem@redhat.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/