Talk:XCB

XLIB/XCB

Most operating systems use Xlib/XCB for their libX11 -- This already has a citation needed mark, but I think it's best just removed from the article. A cursory glance shows that Ubuntu keeps XCB in Universe up through the current release (Natty) (which probably means debian doesn't seem to install it by default), Fedora doesn't have a package(?), and Suse might be installing it by default, I can't quite tell. Additionally, the only window manager that currently uses it (and thus a major reason a distro would install it by default) is Awesome_(window_manager). "An operating system can use Xlib/XCB for its libX11" would be a more correct statement and would share a citation with the next statement. 64.122.73.112 (talk) 16:36, 18 April 2011 (UTC)

Peer review of a related article

I submitted X Window core protocol for peer review, as I intend to candidate it for featured status. I would appreciate comments (Peer review page). - Liberatore(T) 18:07, 21 February 2006 (UTC)

Question

In the example, I see a function XCBWaitForEvent. As far as I understand, this function actually waits for events on the network, and this is different than the corresponding Xlib function (which looks a local queue). If this is correct, it may be a good idea to mention, as it looks like a big differnce. - Liberatore(T) 11:51, 4 May 2006 (UTC)

No, there is also an internal event queue in XCB. Both XCB and Xlib will block if the queue is empty. --IanOsgood 19:33, 24 August 2006 (UTC)

Thanks. This could be added to the article, now or in the future (Liberatore, 2006). 20:52, 24 August 2006 (UTC)

Improved version of XCB example code

This is a matter of opinion, but here is a version that is shorter and takes advantage of C99 features. I think it elucidates the task at hand much more clearly than the existing code. If there are no comments, I will change the article to include this code.

#include <xcb/xcb.h>
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>

int main(int argc, char **argv)
{
	/* open connection with the server */
	xcb_connection_t *c = xcb_connect(NULL,NULL);
	if (xcb_connection_has_error(c))
	{
		errno = ENOTCONN;
		perror(*argv);
		return EXIT_FAILURE;
	}

	/* get the first screen */
	xcb_screen_t *s = xcb_setup_roots_iterator( xcb_get_setup(c) ).data;
	xcb_gcontext_t g = xcb_generate_id(c);
	uint32_t values[2]= {s->black_pixel, 0};
	xcb_create_gc(c, g, s->root, XCB_GC_FOREGROUND | XCB_GC_GRAPHICS_EXPOSURES , values);

	/* create window */
	xcb_window_t w = xcb_generate_id(c);
	values[0] = s->white_pixel;
	values[1] = XCB_EVENT_MASK_EXPOSURE | XCB_EVENT_MASK_KEY_PRESS;
	xcb_create_window(
		c,
		s->root_depth,
		w,
		s->root,
		10, 10, 100, 100, 1,
		XCB_WINDOW_CLASS_INPUT_OUTPUT,
		s->root_visual,
		XCB_CW_BACK_PIXEL | XCB_CW_EVENT_MASK,
		values
	);

	/* map (show) the window */
	xcb_map_window(c, w);
	xcb_flush(c);

	/* event loop */
	xcb_generic_event_t *e;
	for(
		xcb_rectangle_t r = {20,20,60,60};
		( e = xcb_wait_for_event(c) ) &&
			XCB_KEY_PRESS != ( e->response_type & ~0x80 );
		free(e)
	) if(XCB_EXPOSE == (e->response_type & ~0x80))
	{
		/* draw or redraw the window */
		xcb_poly_fill_rectangle(c, w, g, 1, &r);
		xcb_flush(c);
	}
	free(e); // catch the missed free() that occurs when the for-loop exits

	/* close connection to server */
	xcb_disconnect(c);

	return EXIT_SUCCESS;
}

—Preceding unsigned comment added by 216.115.29.1 (talk • contribs) 20:10, 5 April 2007

Well? Should it be changed? --Ysangkok 16:38, 24 April 2007 (UTC)

Having never seen XCB usage before having seen this page, I would give a big fat NO to replacing it with what the anonymous editor has written.

I think the version above is much less readible. Just because C99 allows the things this anonymous editor is doing, such as mixed declarations and code, or declarations within a for(), doesn't mean that these constructs should be used at every opportunity, and it certainly doesn't make the code any clearer that the anonymous contributor has chosen to do so. This is a long function and the anonymous editor has made it much harder to read.

Also, particularly atrocious is the event loop. The anonymous editor has taken an intuitive, concise, standard-issue while/switch event loop, a very common and intuitive idiom, and butchered it into this bizarre illegible for loop followed by a horribly redundant, poorly indented series of if statements. The line of code following the loop also potentially results in freeing a null pointer (should xcb_wait_for_event ever return NULL), which will probably result in a crash for many (most?) libcs.

Seeing the article, I understand what is going on. Seeing the anonymous editor's "cleanup" is another story.

–Andyluciano 19:40, 21 August 2007 (UTC)

No, don't replace the existing code. It is additionally meant to be functionally and structurally similar to the Xlib#Example for purposes of comparison. Also, a lot of free software standardizes on C89 (ANSI C) as a lowest common denominator. --IanOsgood 21:36, 23 August 2007 (UTC)

Whats with the 0x80?

I just read this article and studied the code example carefully. Everything is well written and the example is very readable, except for the line: switch (e->response_type & ~0x80). Shouldn't there be a #define for the ~0x80 whose name conveys some concrete meaning to the corresponding bits of response_type? Or is this relying on some advanced trickery which doesn't make for a good teaching example? Thanks. Jlenthe (talk) 18:30, 10 February 2008 (UTC)

According to this, the higher bit indicates whether the event is synthetic, that is, originating from a sent request of another client. According this explanation, the bit should be removed before checking the event type. And, there is no define for it... Tizio 14:15, 5 September 2008 (UTC)

Documentation?

Anybody knows where to find real documentation on the XCB API? I've searched the XCB website, the Wiki, Google. XcbApi appears to be just a few pages documenting some irrelevant structures and definitions. —Preceding unsigned comment added by Vedge (talk • contribs) 03:11, 7 February 2010 (UTC)

primary sources tag

Every source currently on this topic cites the developers of this library (release announcements by the developers do not count, either). There are no third-party sources cited, no knowledgeable reviews, etc. TEDickey (talk) 10:21, 30 October 2015 (UTC)

Other language bindings

Under "Other language bindings", surely only xcffib should be mentioned as the Python binding. xpyb is obsolete, no longer maintained, only works with Python 2, no longer in most distros' repositories. I propose removing it completely. Any objections? Longitude2 (talk) 21:41, 17 January 2021 (UTC)

xml dialect is not an invention

The topic states that "Creators of XCB have invented a specialized interface description language" which given the trivial example, is some editor's gross exaggeration to claim that an XML-based schema is a "specialized interface description language". TEDickey (talk) 19:07, 22 March 2024 (UTC)