Yes. This would avoid some unnecessarily fragmentation
in the IP packet.
--- Tony Finch <[EMAIL PROTECTED]> wrote:
> Justin Erenkrantz <[EMAIL PROTECTED]> wrote:
> >
> >apr_sendfile doesn't have an implementation in APR
> defined for
> >HP-UX. I do know that HP has different sendfile()
> semantics than
> >Solaris or FreeBSD. I seem to remember Linus
> Torvalds mentioning
> >on lkml that HP-UX had "braindamage" in its
> sendfile implementation:
>
> The problem with sendfile() is the way it is
> frequently used in
> conjunction with write() or writev() -- you use
> write() to send some
> headers and then sendfile() for the bulk data. This
> usually means that
> you get a forced packet boundary between the headers
> and the file
> data, which usually means you aren't making the most
> of the MTU. HP/UX
> addresses this issue by including an iovec in the
> sendfile() call for
> the header; FreeBSD adds a second iovec for the
> trailer. However this
> still leaves you with a forced packet boundary
> between files. Linux
> doesn't include any iovecs in sendfile(); instead
> you must use
> setsockopt(TCP_CORK) to prevent undesired packet
> boundaries.
>
> Tony.
> --
> f.a.n.finch [EMAIL PROTECTED] [EMAIL PROTECTED]
> FAIR ISLE: NORTHWEST 5 INCREASING 6 OR 7,
> OCCASIONALLY GALE 8, BACKING
> SOUTHWEST FOR A TIME. RAIN THEN SHOWERS. MODERATE OR
GOOD.
__________________________________________________
Do You Yahoo!?
Yahoo! Auctions - Buy the things you want at great prices! http://auctions.yahoo.com/
