Linux-2.1.127 (fwd)

[Dean Gaudet <dg...@arctic.org>]

FYI, something I mentioned a while back as a better alternative to turning
off nagle (which causes a packet to be sent for every write()).  I haven't
looked yet to see how you force the kernel to flush the socket. 

Dean

---------- Forwarded message ----------
From: Linus Torvalds <to...@transmeta.com>
[...]
 - various small but important networking fixes from Davem (thanks). One
   of them is the "anti-nagle" bit to allow programs that know what they
   are doing to avoid nagling by telling the kernel so. This is mainly
   things like Web servers and ftp-servers that can use this option
   together with "sendfile()".
[...]

Re: Linux-2.1.127 (fwd)

[Marc Slemko <ma...@worldgate.com>]

When I looked it appears that FreeBSD already has something that does a
similar thing.  There isn't any way to force the push, except I think by
disabling the option. 

It is just a TCP_NOPUSH setsockopt() that was around four lines of code to
add when it was added.

The reason it was added is for T/TCP, so you can buffer a bunch of stuff
before it goes out so T/TCP doesn't require extra segments. 

On Mon, 9 Nov 1998, Dean Gaudet wrote:

> FYI, something I mentioned a while back as a better alternative to turning
> off nagle (which causes a packet to be sent for every write()).  I haven't
> looked yet to see how you force the kernel to flush the socket. 
> 
> Dean
> 
> ---------- Forwarded message ----------
> From: Linus Torvalds <to...@transmeta.com>
> [...]
>  - various small but important networking fixes from Davem (thanks). One
>    of them is the "anti-nagle" bit to allow programs that know what they
>    are doing to avoid nagling by telling the kernel so. This is mainly
>    things like Web servers and ftp-servers that can use this option
>    together with "sendfile()".
> [...]
>