Re: Updates to configure for make-network-process

Re: Updates to configure for make-network-process

[David Ponce]

Hello,

[...]
> I suppose nt/config.nt must be updated as well, but I haven't got a
> clue as to which of the above symbols makes sense there!  Someone
> speaking Windozish, please update...  Thanks.
[...]

I successfully "make bootstrap install" latest CVS version of Emacs 21.3
with the following small patch to nt/config.nt and src/makefile.w32-in.

Hope it will help.
Sincerely,
David

*** nt/config.nt.ori	Thu Mar 14 07:26:58 2002
--- nt/config.nt	Tue Mar 19 09:31:43 2002
***************
*** 242,247 ****
--- 242,256 ----
  /* Standard Windows strftime does not support POSIX.2 extensions.  */
  #define STRFTIME_NO_POSIX2 1
  
+ #define HAVE_SENDTO 1
+ #define HAVE_RECVFROM 1
+ #define HAVE_GETSOCKOPT 1
+ #define HAVE_SETSOCKOPT 1
+ #define HAVE_GETSOCKNAME 1
+ #define HAVE_GETPEERNAME 1
+ 
+ #undef HAVE_SYS_UN_H
+ 
  #define LOCALTIME_CACHE
  #undef HAVE_INET_SOCKETS
  
*** src/makefile.w32-in.ori	Sat Dec 22 23:45:08 2001
--- src/makefile.w32-in	Tue Mar 19 09:36:30 2002
***************
*** 137,142 ****
--- 137,143 ----
  	$(GDI32)	\
  	$(COMDLG32)	\
  	$(USER32)	\
+ 	$(WSOCK32)	\
  	$(MPR)		\
  	$(SHELL32)	\
  	$(libc)




____________________________________________________________
Faites un voeu et puis Voila ! www.voila.fr 
Avec Voila Mail, consultez vos e-mails sur votre mobile Wap. 

Re: Updates to configure for make-network-process

[Jason Rumney]

"David Ponce" <dponce@voila.fr> writes:

> Hello,
> 
> [...]
> > I suppose nt/config.nt must be updated as well, but I haven't got a
> > clue as to which of the above symbols makes sense there!  Someone
> > speaking Windozish, please update...  Thanks.
> [...]
> 
> I successfully "make bootstrap install" latest CVS version of Emacs 21.3
> with the following small patch to nt/config.nt and src/makefile.w32-in.

Did it not build with those constants undefined?  Until that
functionality can be verified on Windows (winsock differs from BSD
sockets in subtle ways), I would prefer to leave these undefined in
CVS.


> + #define HAVE_SENDTO 1
> + #define HAVE_RECVFROM 1
> + #define HAVE_GETSOCKOPT 1
> + #define HAVE_SETSOCKOPT 1
> + #define HAVE_GETSOCKNAME 1
> + #define HAVE_GETPEERNAME 1
> + 
> + #undef HAVE_SYS_UN_H

-- 
Jason Rumney


_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: Updates to configure for make-network-process

[Kim F. Storm]

Jason Rumney <jasonr@gnu.org> writes:

> "David Ponce" <dponce@voila.fr> writes:
> > 
> > I successfully "make bootstrap install" latest CVS version of Emacs 21.3
> > with the following small patch to nt/config.nt and src/makefile.w32-in.
> 
> Did it not build with those constants undefined?  Until that
> functionality can be verified on Windows (winsock differs from BSD
> sockets in subtle ways), I would prefer to leave these undefined in
> CVS.

How would you verify the features without defining them?

Having them defined is not supposed to change anything as long as you
don't use any of the new features of make-network-process.

But defining these now gives those who might be interested a chance to
try out the new features on Windows and provide feedback on whether it
is working - or what might be needed to make it work.

If we don't define them now, when do you propose to define them?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk


_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: Updates to configure for make-network-process

[Jason Rumney]

storm@cua.dk (Kim F. Storm) writes:

> How would you verify the features without defining them?

Define them locally, and fix the problems before checking in. But I
see it is already changed, so it might as well stay like that until
someone gets time to implement wrappers for these functions on Windows.

> Having them defined is not supposed to change anything

Having them defined does change things. It is now necessary to link
against a library that is not present on all versions of Windows,
which previous code has made a lot of effort to accommodate. It is
also necessary to specify extra link options to allow the library to
co-exist with the overridden functions already present in Emacs.


-- 
Jason Rumney


_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: Updates to configure for make-network-process

[Kim F. Storm]

Jason Rumney <jasonr@gnu.org> writes:

> storm@cua.dk (Kim F. Storm) writes:
> 
> > How would you verify the features without defining them?
> 
> Define them locally, and fix the problems before checking in. But I
> see it is already changed, so it might as well stay like that until
> someone gets time to implement wrappers for these functions on Windows.
> 
> > Having them defined is not supposed to change anything
> 
> Having them defined does change things. It is now necessary to link
> against a library that is not present on all versions of Windows,
> which previous code has made a lot of effort to accommodate. It is
> also necessary to specify extra link options to allow the library to
> co-exist with the overridden functions already present in Emacs.

I wasn't aware of this, and the feedback I got didn't indicate
any such problems.

I tried looking at the existing wrappers, and I think I could add the
missing functions -- but I have no way of testing it.

If I understand the code correctly, I have to define the wrappers in w32.c
and add the new functions to nt/inc/socket.h.  I assume I can use e.g.
sys_connect as a skeleton for the new wrappers.  Anything else I need
to be aware of?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk


_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: Updates to configure for make-network-process

[Jason Rumney]

storm@cua.dk (Kim F. Storm) writes:

> I tried looking at the existing wrappers, and I think I could add the
> missing functions -- but I have no way of testing it.

I have started writing the wrappers. Only the wrapper for accept is
non-trivial, and that is the only one I have left to do.  Given that
you don't have a Windows compiler to check that everything compiles, I
don't think it is worth your while trying to add these. Meanwhile, if
anyone tries to use the new functionality on Windows before I check this
in, they will almost certainly not work, and might cause a crash. So
Windows users, please wait a day or two before trying this out.


-- 
Jason Rumney


_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel