Re: [Vala] Memory corruption&coredump on Exit in my first vala program spekle

[Niels Mayer <nielsmayer gmail com>]

On Sat, Aug 28, 2010 at 6:44 PM, Alexander Kojevnikov
<alexander kojevnikov com> wrote:
> You can try running spekle under valgrind [1]. After testing it on a
> flac file I already see that spek_spectrogram_put_pixel() is used with
> wrong arguments causing an out-of-bounds memory write.

Thank you for the clue. I'm especially thankful that I won't have to
run valgrind for hours on end with my CPU fans whining at top-speed
just to decode a video or mp3 :-) ... I will just go straight to
debugging spek_spectrogram_put_pixel().

I've added your suggestion to my issue:
http://code.google.com/p/spekle/issues/detail?id=1#c3

> On a side note, feel free to submit patches adding batch support to
> Spek, there's even an open issue for it [ http://code.google.com/p/spek/issues/detail?id=6 ]

I added the following comment on your issue:
http://code.google.com/p/spek/issues/detail?id=6#c5
...........
http://spekle.googlecode.com is a fork of Spek 0.6 specifically
designed for batch use and for usage without an X server or GUI. It
would be great if the changes for spekle could be incorporated back
into Spek. Although down the road, some functionality I plan to add in
spekle ( http://vamp-plugins.org ) may be out-of-scope for Spek.

It's not something that'll be an obvious patch since I replaced spek.vala with
http://spekle.googlecode.com/svn/trunk/spekle/src/spekle.vala and
and spek-window.vala turned into
http://spekle.googlecode.com/svn/trunk/spekle/src/spek-object.vala and
there's a different set of build targets as well
http://spekle.googlecode.com/svn/trunk/spekle/src/Makefile.am ...
...........
Other code-mangling includes needing to add an extra callback
parameter to the Pipeline() constructor:
http://spekle.googlecode.com/svn/trunk/spekle/src/spek-pipeline.vala
and lots of small changes to
http://spekle.googlecode.com/svn/trunk/spekle/src/spek-spectrogram.vala
. To recognize what changed, comment-outs of your original 'spek-0.6'
code are done with "//" at the beginning of a line, and most of my
changes are marked with an "NPM" comment.

Since I just got 'spekle' running, and my current changes are a
first-pass hacking of your original spek sources, it probably makes
sense for me to continue doing development and debugging on 'spekle'
and at some point in time in the future, submitting a more definitive
patch for http://code.google.com/p/spek/issues/detail?id=6 .

In case you're interested in collaborating on spekle, I added you as a
committer (akojevnikov). It may also make sense to test out batch-mode
ideas in spekle first and then move them back to 'spek' when they've
settled down.

Let me know how you wish to proceed, and thanks for making Spek
available and GPL!

-- Niels
http://nielsmayer.com