[Xcb] Win32 port - pls review

[Jeetu Golani]

Hi,

>>You appear to have accidentally dropped a number of headers in
>>xcb_util.c, including <fcntl.h> and <netinet/tcp.h>, so libxcb doesn't
>>compile any more on *ix.

Ouch...sorry have included them back in the code.

>>#ifndef _WIN32 inside #if USE_POLL seems redundant. USE_POLL 
shouldn't be defined on Win32.

Hmm Win32 doesn't have the poll function and uses WSAPoll instead so 
we'll need to somehow #ifdef the poll function. Pls let me know alternate 
mechanisms to achieve this.

>>The do-while around select in xcb_in.c might not be required on Win32,
>>but does it hurt?

Win32 doesn't have the EINTR error code flag and uses WSAEINTR instead 
so we will have to #ifdef this region or change it somehow. 

Pls let me know what other changes need to be made here and any thoughts 
on the above.

Bye for now
Jeetu
http://www.ebrain.in 
- An open (GPL) platform to discover and run software off someone in the 
meshed network pool and use it on your own device.

twitter:  @0topcat0 and @ebrainpool



On Wednesday 31 March 2010, Peter Harris wrote:
> On 2010-03-30 15:32, Jeetu Golani wrote:
> > I have updated the Win32 port to the current (few weeks old) tree of xcb.
> > Have conducted a few rudimentary tests and it seems to perform all right.
> 
> Looks good. I appear to be down to nits, none of which are critical as
> far as I can see. Thanks for working on this.
> 
> You appear to have accidentally dropped a number of headers in
> xcb_util.c, including <fcntl.h> and <netinet/tcp.h>, so libxcb doesn't
> compile any more on *ix.
> 
> SOCKET on Win32 is currently defined to intptr_t, which is wider than
> int on 64-bit systems. We can probably get away with leaving fd as an
> int, for the sake of keeping the code at least somewhat sane. (Neither
> breaking the ABI on *ix nor #ifdefing the type on Win32 are particularly
> appealing, from what I can see). Note that Win32, unlike posix, does not
> just allocate FDs one by one starting from stderr, so it's more likely
> that some app will run in to this than a *ix person might think. But I
> believe current implementations of Windows try to keep SOCKETs below 
the
> 31-bit barrier most of the time.
> 
> #ifndef _WIN32 inside #if USE_POLL seems redundant. USE_POLL 
shouldn't
> be defined on Win32.
> 
> The do-while around select in xcb_in.c might not be required on Win32,
> but does it hurt? I'd rather avoid as many #ifdefs as possible. In the
> same vein, Remi already suggested using send/recv everywhere (even on
> *ix). What do you think?
> 
> There's trailing whitespace in a few places. "git diff --color" to see
>  them.
> 
> _XCB_WINDEFS_H -- identifiers starting with underscore followed by
> another underscore or a capital letter are reserved in C (7.1.3). I've
> already lost this battle, as the rest of xcb already stomps on this
> namespace, but I thought I'd point it out.
> 
> What does the rest of the list think? Is it better to use a name that
> resembles the rest of the names in XCB, or should we try to avoid the
> reserved namespace going forward?
> 
> Peter Harris
>