Re: [PATCH] move socket-related client parts into separate file

[Derek Robert Price]

Alexey Mahotkin wrote:

> > > > > > "DRP" == Derek Robert Price <derek@ximbiot.com> writes:
> > > > > >            
>
> DRP> Which part is lose-lose?  Supporting Windows?
>
> I'm just scared that there are 300 lines which in principle could not be,
> e.g. by switching Unix to recv() too.  Ok.
>
> Oh!  Maybe should just get rid of #ifdef NO_SOCKET_TO_FD in
> socket-client.c, so that it is always compile-checked, but only use
> resulting buffer under Windows?  Let's think about it later.

Actually, I'd rather make the whole file not compile at all at the 
Makefile level when NO_SOCKET_TO_FD is defined.  That way executables 
are smaller.  I started looking into it and decided I don't have time at 
the moment, though.

As for compiling and testing it on UNIX, it is probably sufficient to do 
that only before committing changes to that part of the code.  I'll get 
the nightly testing on Windows up eventually.  :)

> DRP> Committed, on top of Alexey's previous socket-client patch.
>
> I just yesterday worked on refreshing all of this :) Ok, I'm sending
> from_/to_server set of patches now.  Please consider.
>
> In the evening I'll refresh logging buffer also, and one more patch to
> socket-client (init_sockaddr() needs to be moved there too, to be used by
> GSSAPI/Kerberos).
>  

I just committed your rsh-client patch as well.  Does this mean you want 
me to wait before committing your other six patches?

Derek

--
               *8^)

Email: derek@ximbiot.com

Get CVS support at <http://ximbiot.com>!
--
I will not do that thing with my tongue.
I will not do that thing with my tongue.
I will not do that thing with my tongue...

         - Bart Simpson on chalkboard, _The Simpsons_