[Xcb] char* in structs ;)

[Vincent Torri]

On Wed, 15 Jun 2005, Jamey Sharp wrote:

> On Wed, 2005-06-15 at 13:08 -0400, Jeremy Kolb wrote:
> > Quoting Jamey Sharp <jamey at minilop.net>:
> > > On Wed, 2005-06-15 at 08:26 -0400, Jeremy Kolb wrote:
> > > > Well the user can provide the display with the standard -display paramter
> > > (:1,
> > > > host:display ...)  I'm just not sure what the functions are to make use
> > > of
> > > > that.
> > >
> > > Oh, OK. I consider that outside the scope of XCB: it won't save your
> > > display string for you. You should save it yourself if you want it.
> > > Also, obviously, XCB doesn't do command-line argument processing for you
> > > either. :-)
> >
> > Right I know that, what I'm asking is... if I have the string, how do I
> > generate the appropriate connection.  I believe in xlib you can do dpy=
> > XOpenDisplay(display_string).
> >
> > Or something like that, I don't have any code in front of me and I don't know
> > xlib that well so I may be lying.  Is this sort of thing outside of the scope
> > of XCB or am I blind (very distinct possibility ;)).
>
> Oh, gosh. OK, yes, obviously that is part of XCB's job. I completely
> misunderstood your question.
>
> The API around this is scheduled to change, though I'm not 100%
> satisfied with the current plan for the new API, which is part of why
> it's not there yet. (The other part is that I'm either lazy or busy,
> depending on how generous you want to be.)
>
> In the current API, you can use this code (completely untested, but I
> think it's right):
>
>         char *host;
>         int display;
>         int fd;
>         XCBAuthInfo auth = { 0 };
>         XCBConnection *c;
>
>         XCBParseDisplay(display_string, &host, &display, 0);
>         fd = XCBOpen(host, display);
>         XCBGetAuthInfo(fd, &auth);
>         c = XCBConnect(fd, &auth);
>
> Awful, ain't it? And I omitted error-checking for clarity:
> XCBParseDisplay returns 0 on failure, XCBOpen returns -1, and XCBConnect
> returns a null pointer. XCBGetAuthInfo can fail too but if it does you
> can (and probably should) still try to connect.
>
> In the currently planned API, you'll do this:
>
>         XCBConnect(display_string, 0);
>
> This incompatible API change doesn't really seem worth doing, does it?


Jeremy,

Horror !! You haven't read my tutorial !!! It's in it !!! Shame on you !

Vincent