2007/12/30, IOhannes m zmoelnig <zmoelnig@iem.at>:

Olivier Heinry wrote:
> Hi,
>
> there's a problem with the pix_record object which is atm makes it
> totally unusable on debian testing. Unofortunately for me , I did a
> patch which relies on it, and that used to work until May , and when I
> reopened it in December dont work any more. Neither the official Debian
> package of Gem nor Pd-extended or CVS compiled version work. I asked

that is very weird, as afair at least the debian package doesn't even
feature the [pix_record] object.
if it does, then this gets even weirder, as the debian package has not
been updated for a long time (it is still the last official 0.90 release)
the only thing that has changed on the debian side is quicktime
(upgraded from libquicktime0 to libquicktime1), so it might be related
to that.

Well, I just reinstalled a debian stable instead of lenny (too much maintainance) , installed Pd-extended 0.39-3 from the Sourceforge link and pix_record works again. (CVS version compiled 2007 Oct. 21st).

> So I guess it is related to the source code (i read the archive of
> gem-dev about hard coded 20fps and the many pd crashes that occur ) and

the hardcoding of 20fps should do nothing to the stability of the object.
the only drawback of this is, that the recorded movie will be played
back (by other media-players than Gem; which respect the framerate) at
the wrong speed (too slow, too fast)

OK.

The really annoying bug is the following: if you use pix_record as a top-level object such as in the help file.
Once you nest it into an abstraction and/or subpatch, it works the first time you record, crashes the second time.

It behaves this way on my x86 Debian Etch box,
libquicktime is libquicktime0 (0.9.7-1)
Benjamin Cadon told me the same happened to him
(Benjamin, could you post your details?) .

I gave a try to Pure:Dyne's Gem version
pix_record from puredyne-2.3.66-DVD-RC1 and before
beahves the same as described before
pix_record from puredyne-2.3.66-DVD-RC2 is broken (object doesnt create at all)

> I was wondering if there was a way have its bugs corrected (be it by the
> magical powers of a bounty) or will it stay postponed to the Greek
> calendas?

- file a bug-report (please file it at pd-gem's tracker not at the
pure-data tracker)

>
> Is it more reasonable to patch a woraround with pix_write?

not very reasonable (though working)

i would prefer having [pix_record] fixed.

me too! One workaround that benjamin did is to reinstatiaate the abstraction after each recording

++
O.