[PATCH] Counting I/O with FLUSH w/o MMN bump

[Bojan Smojver <bo...@rexursive.com>]

Although the subject claims no MMN bump and the patch doesn't have one, I'm not
sure if that's at all realistic. The patch does compile and it does work, so at
least something is good about it :-)

Give it a whirl, tell me what you think and where do we go from here...

Bojan

Re: [PATCH] Counting I/O with FLUSH w/o MMN bump

[Bojan Smojver <bo...@rexursive.com>]

Quoting Bojan Smojver <bo...@rexursive.com>:

> The apr_socket_t *csd is the first field of the core_conn_conf
> structure, so old modules that pick it up as a single pointer should
> still find the thing in exactly the same place. Unless the new alignment
> (which should be the same as before) or something I'm not aware of
> screwed things up. And because core_conn_conf is brand new, no old
> module should be stuffing anything beyond apr_socket_t* either...

Actually, I think I told a lie here. The old code was storing a pointer to
apr_socket_t into the core module config. The new code stores the pointer to the
core_conn_conf into the core module config, not the core_conn_conf itself. Which
means that if there are any third party modules out there that depend on the
pointer to apr_socket_t to be in core module config, they will be broken for sure.

Bummer :-(

Bojan

Re: [PATCH] Counting I/O with FLUSH w/o MMN bump

[Bojan Smojver <bo...@rexursive.com>]

On Tue, 2002-10-22 at 17:50, Bojan Smojver wrote:
> Although the subject claims no MMN bump and the patch doesn't have one, I'm not
> sure if that's at all realistic.

Just one clarification on the "realistic" business...

The apr_socket_t *csd is the first field of the core_conn_conf
structure, so old modules that pick it up as a single pointer should
still find the thing in exactly the same place. Unless the new alignment
(which should be the same as before) or something I'm not aware of
screwed things up. And because core_conn_conf is brand new, no old
module should be stuffing anything beyond apr_socket_t* either...

However, reality has been known to prove me wrong :-)

Bojan