[Xcb] Win32 port -- git help

[Jeetu Golani]

Hi,

I've updated the current (about a couple of weeks ago) code of xcb to include 
changes for Win32. I'm still testing this code however it seems all right at 
the moment.

What I would like to do is push this new branch onto my github repo which 
would then contain the code for the earlier version and the updated (and 
history).

Unfortunately I'm not very good with git :(....I wonder what is the preferred 
practice by xcb devels to do this. Should I establish a new branch at my 
github repo? How should I tag this new branch? Should I then merge this new 
branch to the current tree there? 

I'd appreciate advice (and git steps :) ) to achieve this updation.

Bye for now


On Wednesday 24 February 2010, Jeetu Golani wrote:
> Hello Jamey,
> 
> >> Why does your patch add an initWSA function to the Windows builds of
> >> XCB?
> 
> I thought I had taken care of this so that WSAStartup and WSACleanup are
> handled by the caller. Will look and see if I'm not doing this. If you
>  found this as part of the old patches then that's obsolete work and if I'm
>  not mistaken the code in github shouldn't have this.
> 
> >> It's a small thing, but please be careful about matching the code
> >> style of the rest of XCB's source.
> 
> Probably my bad :) will work this out in my next update.
> 
> >> We don't printf in libxcb, or if we do it's a mistake.
> 
> Aargh I thought I'd got rid of this too. Need to review my code :) Will
>  take it off if it's still there.
> 
> >> One way you can cut that down
> >> some is that I think we can replace read and write with recv and send
> >> on Unix systems just as well as you can on Windows. Then the major
> >> differences seem to be errno vs. WSAGetLastError(), and 
EWOULDBLOCK
> >> vs. WSAEWOULDBLOCK.
> 
> Yup this should prevent duplication.
> 
> >> I'd really like other folks to look them over, especially developers on
> 
> other non-Linux platforms
> 
> >> that we currently support.
> 
> I second that, would be nice hearing from other non-Linux platforms to see
> if this breaks anything out there.
> 
> >> Jeetu Tweeted an "ack" to this request and I replied with a promise to
> >> actually review it this time. :-)
> :
> :)...hopefully will have this sorted out soon....anyone else on this list
> : on
> 
> twitter??
> 
> 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 Wed, Feb 24, 2010 at 1:32 PM, Jamey Sharp <jamey at minilop.net> 
wrote:
> > On Tue, Feb 23, 2010 at 1:27 AM, Barton C Massey <bart at cs.pdx.edu> 
wrote:
> > > ... I'll merge it myself [...] unless
> > > somebody speaks right up right now with a *really* good
> > > reason not to.
> >
> > I do still want review on the changes. I'd really like other folks to
> > look them over, especially developers on other non-Linux platforms
> > that we currently support.
> >
> > I've gone ahead and reviewed the old patches a little, so I have some
> > comments now... (Yes, apparently it's possible to shame me into doing
> > patch review when I'm doing badly at it.)
> >
> > Most importantly: There's nothing here that makes me think we can't
> > support Windows in XCB. I have some issues with style and other
> > details in places, but the fundamental changes are fine.
> >
> > Why does your patch add an initWSA function to the Windows builds of
> > XCB? I thought it's the caller's responsibility to call WSAStartup and
> > WSACleanup? I don't know how that's normally handled in Windows
> > libraries, but at the least, the name "initWSA" doesn't fit XCB's
> > public naming conventions. Or maybe you meant to put initWSA in
> > xcbint.h rather than xcb.h, so applications can't call it?
> >
> > It's a small thing, but please be careful about matching the code
> > style of the rest of XCB's source. github may be misleading me here,
> > but the whitespace looks pretty inconsistent. Maybe github handles
> > tabs poorly? But we don't use tabs for indentation in XCB anyway. (I
> > would pick a different code style if I started again, but it hardly
> > seems worth changing...)
> >
> > We don't printf in libxcb, or if we do it's a mistake. Someday
> > _xcb_conn_shutdown will probably allow setting a more detailed error,
> > but for now just drop the debugging printf calls.
> >
> > I'd feel a lot better about merging this if it didn't involve
> > copy/pasting some good-sized functions. One way you can cut that down
> > some is that I think we can replace read and write with recv and send
> > on Unix systems just as well as you can on Windows. Then the major
> > differences seem to be errno vs. WSAGetLastError(), and 
EWOULDBLOCK
> > vs. WSAEWOULDBLOCK.
> >
> > This work is clearly worth cleaning up and merging. I'm sorry we
> > haven't done it before now.
> >
> > > Maybe you can bring the port up to date one more time?
> >
> > Jeetu Tweeted an "ack" to this request and I replied with a promise to
> > actually review it this time. :-)
> >
> > Jamey
>