Re: [Qtractor-devel] Regressions in current drumkv1 Git

[Rui N. C. <rn...@rn...>]

On 10/7/18 8:14 PM, Frank Neumann wrote:
> 
> Hi Rui and all,
> 
>>>> $ git log --oneline
>>>> 7778e0d - Yet some more offset sync state fixing. <-- gives the reported issue above
>>>> a2114b6 Yet another round to offset sync fixing.
>>>> a25be50 - Fix LV2 UI initialization (again).      <-- this version still works fine
>>>> ...
>>>>
>>>> Any idea what is going on here?
>>>> My qtractor version is 0.9.3, commit ID af83cd.
>>>>
>>>
>>> hi Frank,
>>>
>>> confirmed: current git head (7778e0d) is bonkers :)
>>>
>>> will try to fix it asap.
>>>
>>
>> maybe fixed now, hopefully! :)
>>
>> drumkv1.git $ git log --oneline
>> 2c5cbd2 (HEAD) - Again some more reverse and offset sync state fixing.
>> 7778e0d - Yet some more offset sync state fixing.
>> a2114b6 Yet another round to offset sync fixing.
> 
> 
> Yep, looking good - great reaction time again :-). Thanks a lot!
> 
> 
> On a sidenote..I was trying the last few days to change the loaded preset file
> in drumkv1 while it is actively playing back a short (looped) MIDI pattern
> (this is for e.g. auditing different drumkits while the underlying pattern
> is playing and unchanged).
> 
> This works fine *most* of the time, but every now and then I can make it to
> crash. I was wondering - should drumkv1 be basically robust to this, or can
> it happen it attempts to e.g. play back a sample that was free()d just a ms
> before?
> 
> I rebuilt drumkv1 with --enable-debug and could produce a core dump and gdb
> backtrace. It looks like the following; is that enough to start debugging?
> 

don't bother with that

loading samples let alone a set of samples while it's playing some is
not a safe operation. please refrain and avoid doing that unless you
enjoy the fun of fireworks burning in your face :)

cheers
-- 
rncbc aka. Rui Nuno Capela
rn...@rn...