[PD] pix_snap + pix_record = problem

[Matteo Sisti Sette]

Hi,

I think I've found what triggers the problem (and avoiding it I can get 
things to work fine).

It seems to happen when these things happen in this order:

1) The [gemhead] driving the [pix_snap] and [pix_record] is disabled 
(via a 0 message) _before_ the gemwin is created (for example, at loadbang)

2) After creating the gemwin, the gemhead is not enabled until _just 
before_ the "record 1" and "auto 1" messages are sent to [pix_record].

That is, the following sequence of actions, that should be perfectly 
legal, leads to malfunctioning:

i) disabling the gemhead driving pix_snap and pix_record
(some time passes)
ii) creating the gemwin
(some time passes)
iii) filename is given to pix_record
(some time passes)
iv) the following is done in 0 logical time:
iv.1) enabling the gemhead
iv.2) "record 1" message to pix_record
iv.3) "auto 1" message to pix_record


This can lead to either of two malfunctioning and unrecoverable situations:

A) immediate crash

B) pix_record outputs some error messages and does not record;
    The pix_record will never recover. Even if you do a sequence of 
steps that would normally lead to good functioning (e.g.: 
destroy/recreate the gemwin, enabling the gemhead, repeating steps iii, 
iv.2 and iv.3 in non-zero time), it won't work any more.


I wouldn't expect this to cause any problem; at the very least, 
pix_record should:
- never crash
- give a meaningful error message if the situation presented is actually 
illegal for some reason
- recover and work correctly when "normality" is reestablished


Well all this is basically a bug report so I will file it to the bug 
tracker. I just thoght I'd let you know that I had found the workaround; 
then while writing it came out as if i was filing a bug report...
Also, if anybody can confirm the issue on other platforms beside windows 
and/or explain why this behaviour should be expected (except the crashes 
of course), that would be helpful.

thanks
m.



news at petervenus.de escribió:
> Hello
> I am working with [pix_snap] and its running quite good.
> In the attachment you will find a test-patch i build out of help-files.
> While playing around with [pix_snap]i noticed, that its quite important, 
> to have different gemheads for the parts where you do something
>  to the source-signal and the part of the patch where you record the video.
> like having [gemhead 5] for the first, and [gemhead 55] for the snapping part.
> that way the first part gets handled before the image is snapped.
> hope that helps.
> 
> Peter
> 
> -----Ursprüngliche Nachricht-----
> Von: Matteo Sisti Sette 
> Gesendet: 21.03.2010 13:23:16
> An: Jaime Oliver 
> Betreff: Re: [PD] pix_snap + pix_record = problem
> 
> Jaime Oliver escribió:
>> I had a similar error in osx10.6. 
> 
> Did you find a workaround? Could you ever record a sequence of snapped 
> pixes into a video file?
> 
> thanks
> m.
> 
> 
> 


-- 
Matteo Sisti Sette
matteosistisette at gmail.com
http://www.matteosistisette.com
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: pix_snap_test_bug.pd
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20100321/2f7b4ac1/attachment.asc>