AllowEmptyInput and HAL

[Phil Endecott]

Peter Hutterer wrote:
> On Thu, May 07, 2009 at 11:50:40AM +0100, Phil Endecott wrote:
>> (gdb) print dev->key->xkbInfo.state
>> $4 = {group = 0 '\0', base_group = 0, latched_group = 0, locked_group =  
>> 0 '\0', mods = 0 '\0', base_mods = 0 '\0', latched_mods = 0 '\0',  
>> locked_mods = 0 '\0',
>>   compat_state = 0 '\0', grab_mods = 0 '\0', compat_grab_mods = 0 '\0',  
>> lookup_mods = 0 '\0', compat_lookup_mods = 0 '\0', ptr_buttons = 0}
>>
>> Here initially mods=0, but it changes:
>>
>> (gdb) cont
>> Continuing.
>>
>> Breakpoint 1, XkbHandleActions (dev=0x82bf838, kbd=0x82bf838,  
>> event=0x82c8268) at xkbActions.c:1072
>> 1072	{
>> (gdb) print dev->key->xkbInfo.state
>> $7 = {group = 0 '\0', base_group = 0, latched_group = 0, locked_group =  
>> 0 '\0', mods = 5 '\005', base_mods = 5 '\005', latched_mods = 0 '\0',  
>> locked_mods = 5 '\005',
>>   compat_state = 5 '\005', grab_mods = 5 '\005', compat_grab_mods = 5  
>> '\005', lookup_mods = 5 '\005', compat_lookup_mods = 5 '\005',  
>> ptr_buttons = 0}
>>
>> I see a pattern of mods=0 and mods=5 that's consistent with the  
>> alternating pattern that I posted in the xev output previously.
>
> this is... weird. does it happen after you run 
> setxkbmap -layout "us" -model "pc105" -rules "evdev" -option "" -variant ""
> (just to exclude potentially broken layouts)

No change.  (And that was quite hard to type with every other character 
being interpreted as a control key!)

I don't think I have time to debug this any more at this level 
(especially as there seems to be a 24-hour turnaround here; presumably 
we're on opposite sides of the world).  I'm going to try to purge 
everything X- or HAL-related and install the current Debian packages.  
Please cross your fingers for me...

Phil.