Xboard sluggish on OSX

[JoshPettus]

I just tried the second board observing a game of bughouse. The new resizing is very much an improvement, and works great when shrinking the board. When expanding however, the window expands, but the redrawn board only fills the space of the old size and you only see a portion of the board. Only when you resize the second window does it recacluate the draw size to be the size of the main board.

[hgm]

Hmm, I though I had fixed that. Apparently not reliably...

The problem is that the expose event that changes the size of the secondary window somehow does not refresh the canvas to the new size of the window.

[JoshPettus]

I see the mailing list item you posted. Unfortunately, or fortunately, apple doesn't maintain their version of XQuartz anymore and leave that up to the XQuartz project. Hopefully they are aware, I'll check.

[EDIT] I checked their bugzilla, It doesn't look like it. I'm not sure I'm aware enough of the problem to make a useful bug report outside Your version of X server is slow, fix it. :-/

[hgm]

Can you put the following printf statement in GraphEventProc() in gtk/xoptions.c:

Code: Select all

	    if(sizing) { // the memory buffer was already created in GenericPopup(),
			 // to give drawing routines opportunity to use it before first expose event
			 // (which are only processed when main gets to the event loop, so after all init!)
			 // so only change when size is no longer good
		cairo_t *cr;
printf("new canvas %x %dx%d\n",graph,w,h);
		if(graph->choice) cairo_surface_destroy((cairo_surface_t *) graph->choice);

and then observe two ICS games, and enlarge the main window to trigger this error, to see what it prints?

Unfortunately it will print a lot, almost all from the main window (recognizable by a different hexadecimal number for the graph handle), as every mouse motion event will trigger this code. But the important thing is that once you are done sizing, you should also get an output for the other window.

[JoshPettus]

Can you put the following printf statement in GraphEventProc() in gtk/xoptions.c:

Here is what I got, hope it helps

Code: Select all

fics% new canvas d1438 485x485
new canvas d1438 485x485
new canvas d1438 485x485
new canvas d1438 485x485
new canvas d1438 485x485
new canvas d11e0 539x470
new canvas d11e0 634x554
new canvas d11e0 720x672
new canvas d11e0 727x683
new canvas d11e0 727x682
new canvas d1438 485x485
new canvas d11e0 681x681
new canvas d1438 485x485

[JuLieN]

H.G., IIRC there had been discussions, a while ago, regarding a work-in-progress Cocoa version of XBoard. Has it been canceled ?

[JoshPettus]

Is this regarding Voice Over support?

That was MaddtheSane's project, and unfortunately he hasn't touched it since 2013 which was what brought me to look into GTK on OSX.

https://github.com/MaddTheSane/MacBoard

[JuLieN]

Is this regarding Voice Over support?

That was MaddtheSane's project, and unfortunately he hasn't touched it since 2013 which was what brought me to look into GTK on OSX.

https://github.com/MaddTheSane/MacBoard

No, I'm talking about Chris Sears' OSXBoard port :
http://www.open-aurec.com/wbforum/viewt ... =2&t=52170

[JoshPettus]

Wow I thought it was the same guy. Guess not, Madd's name is C.W. Betts
Haven't heard from either in a very long time.

[hgm]

Well, at some point Chris Sears stopped answering my e-mails.