[PD-dev] Re: [GEM-dev] Funny crash in pd/Gem

chris clepper cgc at humboldtblvd.com
Wed Aug 24 04:54:49 CEST 2005


I will commit the fix in the next two days.  There is a small issue  
with pix_coordinate that I may have resolved but needs a bit more  
testing.

cgc


On Aug 23, 2005, at 11:45 AM, B. Bogart wrote:

> Ah, I now remember chris's post.
>
> Chris, could you send me a patch or commit your fix?
>
> I have an example patch that indeed triggers some mess using  
> separator,
> since I don't see the crash without it. (well at least after an  
> hour so far)
>
> Thanks Erich,
>
> B>
>
> Erich Berger wrote:
>
>> hi ben,
>>
>> i experienced something similar a while ago,
>> and what i found out was that pd/gem slowly eats
>> up memory after time.
>> as i can remember wasnt there a few weeks ago something
>> about a memory leak in "separator" ?
>>
>> have a look at your memory over time,
>> maybe its related ...
>>
>> erich
>>
>>
>>
>>
>> ---------------------
>> http://randomseed.org
>>
>> On Tue, 23 Aug 2005, B. Bogart wrote:
>>
>>
>>> Hey all,
>>>
>>> After running my patch for about 2-3 hours it will suddently  
>>> exit. This
>>> could be related to some off issues I've been having with CPU  
>>> usage, PD
>>> using up 90% when it should be using more like 20%.
>>>
>>> What confuses me is that all gdb returns is:
>>>
>>> pd_gui: pd process exited
>>> Couldn't get registers: No such process.
>>> (gdb) where
>>> Cannot fetch general-purpose registers for thread 1076267520:  
>>> generic
>>> error
>>>
>>> Is this a watchdog kill?
>>>
>>> Anyhow I'll try and pinpoint the cpu usage issue and see if that  
>>> is the
>>> cause.
>>>
>>> I'm running pd 0.38-4 on debian/sarge using the demudi packages.
>>>
>>> Thanks,
>>> B>
>>>
>>>
>>
>> _______________________________________________
>> PD-dev mailing list
>> PD-dev at iem.at
>> http://lists.puredata.info/listinfo/pd-dev
>>
>>
> _______________________________________________
> PD-dev mailing list
> PD-dev at iem.at
> http://lists.puredata.info/listinfo/pd-dev
>





More information about the Pd-dev mailing list