On May 11, 2010, at 12:58 AM, Thomas Adam wrote:

> On 11 May 2010 00:34, Omar Zakaria <[email protected]> wrote:
>> Hey everyone. I'm seeing a strange problem with FVWM v2.5.27. I'm
>> not sure if it's a bug or some odd configuration on my end or what,
>> but I'd appreciate some help all the same.
> 
> Use 2.5.30, please.

Thanks for getting back to me so quickly, Tom.

Just compiled and installed it now. I'm afraid there's no change. Actually,
I now have to kill the program twice before I can open windows correctly.

I just went back and confirmed that this has always been the case: On
first launch, I only get to the app's splash screen before it hangs. On 
the second launch, I get to the app's main window, and on the third
launch, I can use the app freely.

> 
>> So, I've got 2.5.27 installed on a Redhat EL5.5 machine and in
>> general, it seems to work fine. We also have installed on the
>> machine an in-house Smalltalk app that runs using Cincom
>> Visualworks 7. When we open the app, the first window
>> appears just fine, but if we try to open any other of the
>> application's windows, the whole thing freezes, the new window
>> does not appear, and we have to forcibly kill the process. Then,
>> even stranger, the next time we load up the app, everything
>> works fine and we can open as many windows are our hearts
>> desire. But, when we reboot, log out, or otherwise end our login
>> session, we have to repeat the process over again (i.e. launch
>>  the app, close it, and launch it again before we can actually use
>> it.)
> 
> What is this program written in?  Smalltalk you say?

Yes.
In case you're not familiar with it, it's a language that runs in a VM.
The premise behind it is very similar to how Java works: you have
to provide the VM with an image file -- the Smalltalk equivalent of
java's jar files -- which the user creates. The VM then executes the
image file.

We use Cincom's VisualWorks VM version 7.3.

> 
>> I launched the smalltalk app in gdb and found that it was hanging
>> in an Xlib call: XPeekIfEvent(). I can't really introspect it more than
>> that, I'm afraid. Strangely, when I launched the app in gdb again
>> afterwards (the second time, so everything should be cool) and
>> asked it to break on an XPeekIfEvent() call, I found that no such
>> call was made.
> 
> Can you get a proper stack trace, please?

Aye aye. I'm afraid it's not particularly helpful, but here it is:

#0  0xb7f72402 in __kernel_vsyscall ()
#1  0x008a088d in ___newselect_nocancel () from /lib/libc.so.6
#2  0x009861d5 in ?? () from /usr/lib/libxcb.so.1
#3  0x00987bac in xcb_wait_for_event () from /usr/lib/libxcb.so.1
#4  0x009e58f1 in ?? () from /usr/lib/libX11.so.6
#5  0x009e5c17 in ?? () from /usr/lib/libX11.so.6
#6  0x009cf5ee in XPeekIfEvent () from /usr/lib/libX11.so.6
#7  0x080cf438 in ?? ()
#8  0x099b0ba0 in ?? ()
#9  0xbf928900 in ?? ()
#10 0x080cf360 in ?? ()
#11 0xbf928984 in ?? ()
#12 0x00000000 in ?? ()

I suppose it could be our version of libX11 that's problematic... we're
using libX11 from an rpm: libX11-1.1.3-4.el5.

> 
>> Does anyone have any ideas as to what's going on or whether
>> this is indeed a bug?
> 
> Is there a way for me to get ahold of this application so I can see if
> I can reproduce it myself?  Or can you find some other application
> which does the same thing?  It could be *anything* at this point --
> maybe some of the windows are still in the Withdrawn state when
> they're mapped, etc.,e tc., but that's just pure speculation, and I've
> no way of verifying that.

I'll see if I can get authorization to send you the image. 

You might first download the Cincom VisualWorks VM. Like I said,
we're using v7.3.1. I think you can download that particular version 
from here: http://cincomsmalltalk.com/scripts/DownloadCD.ssp.

I'll see about getting something else to act the same way. At the
very least, I might be able to whip up a simple Smalltalk app that
does the same thing.

Thanks again.

--
Omar Zakaria
Agilent Technologies
W: 707-577-4214

Reply via email to