[ggo-discussion] Re: gGo error/request

Hello,

sorry for the delay of my reply, been busy this week with other things and
now I'm slowly digging myself through that pile of gGo emails which stacked
up.


On Sun, Mar 30, 2003 at 08:33:06PM -0500, Andrew D. Esther wrote:

> Bug: I have had this problem about 3-4 times. If you close down a observed =
> game right after you accept a match and before your new match comes up an=
> d if the other person is black. I cannot see the first move. It will even=
> tually say that I timed out. (I can't move because opponent hasnt moved, =
> so I cannot stop it either). I have tried 'refresh' in the command window=
> n and that did not fix it either. Perhaps now you can recreate the bug it=
>  to figure it out.

I know this, this is actually an old problem, but I failed to fix it
properly so far. Blame my incompetence for it. :)

This problem occurs in two variations:
* As you write, when you play white, then game opens and you dont see the
black move.
* Sometimes when observing, new moves would stop being displayed on the
board.

Some of my observations concerning both, which seem to be based on the same
original problem:
* It happens only on Windows, not Linux or OS X. Please correct me if you
got it on a non-Windows platform.
* It happens absolutely randomly. I fail to reproduce it reliably.
* It seems to be a matter of focus problem. When you switch the focus within
the gGo Windows (mouse click or alt-tab), that might trigger the problem. It
usually does not appear if you dont touch the mouse. (Of course, when a new
game opens, a focus change is involved).
* The stones *are* painted on the board (!), but not displayed. You can
confirm this by either resizing the window or minimize/restore it, which
will trigger the operating system to redraw the window, and then the
previously missing stones appearand things continue normally. So somehow the
window stops updating itself, for whatever reason, and adds only invisible
stones.

I suppose, but am not sure, latter hints to the problem. I enabled some
extensive debug messages, and those moves which do not get displayed
actually run through all the methods in the code properly, including the
method which really paints them on the board. The method which calls the
update of the window is also called.

So far, the process-incoming-move-and-display-it-on-the-board within gGo
seems to work properly. Also, the stones are painted on the board. So far we
talk about gGo and Java code. But now the window sometimes does not update.
And I think we now talk about Java and operating system code, or rather
integration of Java with the OS libraries. The observation the problem only
occurs on Windows would support this theory, also the observation that the
problem can be somewhat triggered, although unreliable, when switching the
window focus seems to support it.

Well, I write much but finally I need to summarize it with: I've no clue, it
seems to work properly when I look at the debug output, and I never got the
problem on Linux (on which I develop gGo).

After all, this is one of the oldest and most mysterious persistant gGo
bugs. Allright, other software like CGoban2 and Jago dont have it, so of
course it must be a problem in my program, no doubt on that.


> Request: I also have a request to add a sound to when people enter/leave a =
> channel, (turn channel sound on and off in preferences). This is alert me=
>  when friends join my channel. I do not like to find out that I missed a =
> friend because I had my channel downsized...

Possible, but not for the 0.3.5 release (which should be available today).


 Peter

Other related posts: