-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Am 01.02.2011 23:32, Ng Oon-Ee wrote:
> On Tue, 2011-02-01 at 21:07 +0100, Uli Schlachter wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA256
>>
>> Am 01.02.2011 08:39, Ng Oon-Ee wrote:
>>> After some more testing (with other apps as well), it seems the
>>> 'offending' hint is this one (from xprop):-
>>>
>>> WM_TRANSIENT_FOR(WINDOW): window id # 0x5a00001
>>>
>>> How does awesome handle this particular hint/
>>
>> I can't find anything in the C core actually using this, except for the 
>> stacking
>> order (client's are above the window for which they are transient).
>> It's available to lua via c.transient_for. The only thing that uses this is
>> awful.tag. When a new client with a transient_for hint appears, it gets the 
>> same
>> screen and tags as its transient for client.
>>
>> So, that shouldn't have anything to do with this.
>>
>> Cheers,
>> Uli
> 
> Hmm, just retested, while yesterday I could not find the parent window
> today I can. Here's xwininfo -id 0x5600001 (which is the ID currently
> under WM_TRANSIENT_FOR for the app.
> 
> xwininfo: Window id: 0x5600001 "e-Sword"
> 
>   Absolute upper-left X:  640
>   Absolute upper-left Y:  400
>   Relative upper-left X:  640
>   Relative upper-left Y:  400
>   Width: 1
>   Height: 1
>   Depth: 24
>   Visual: 0x21
>   Visual Class: TrueColor
>   Border width: 1
>   Class: InputOutput
>   Colormap: 0x4800001 (not installed)
>   Bit Gravity State: NorthWestGravity
>   Window Gravity State: NorthWestGravity
>   Backing Store State: NotUseful
>   Save Under State: no
>   Map State: IsUnMapped
>   Override Redirect State: no
>   Corners:  +640+400  -1917+400  -1917-621  +640-621
>   -geometry 1x1+640+400
> 
> So it seems there's an unmapped parent window with a 1x1 pixel size. How
> would this affect tiling?

This is IMHO an ICCCM violation which states[1]:

"Clients should not rely on transient windows being available to the user when
the transient owner window is not in the Normal state."

The transient owner window is unmapped which means that awesome isn't managing
it which in turn means that it is in the "withdrawn" state. From my reading of
ICCCM this means that awesome is free to hide that e-Sword window.


Now for what actually happens: I haven't verified yet, but it looks like this
case triggers a bug and awesome aborts half-way through making that window
visible. Could you check if you get any error messages on awesome standard error
output? E.g. on Linux:

  ls -l /proc/$(pidof awesome)/fd/2

If that points to some file, error messages end up in that file. If this points
to e.g. "tty1" then error messages are sent to the tty where awesome was 
started.

Cheers,
Uli

[1] http://tronche.com/gui/x/icccm/sec-4.html#s-4.1.4

- -- 
The Angels have the phone box!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)

iQEcBAEBCAAGBQJNSWi8AAoJECLkKOvLj8sGj7oH/1WYyUjBtw2qDlM1aZ+PZgrJ
7DMrVciH45UjzoYg4TGJw4W8ahGnrOvf4e4q4BvGlM6eMcK0cfV7VhLHsEwjmKn6
Og29psLS+G60bZURSzUNTb5/Nf1kyMSwQCd2jO8+/P7nMDmqIBp7UYQQw2WQ4wpW
4Cm12fqjcslKakYvU/O8T0PQk+aHczASUliVUorDeYSBfA2HLjOf87ovmjsiMeqN
wA4ignYNDPknX3cN6mOsgRbmzzrIwCRSK7iuofnXtQB1SOunnuLAuuSOQMyTLX5i
eTck0ypWnzyG2hfpA3nKRBFEMok+b2lWwLbKdbxRs+74YEXolqxsS76NV1AiYaQ=
=/1lz
-----END PGP SIGNATURE-----

-- 
To unsubscribe, send mail to [email protected].

Reply via email to