Tim Blechmann wrote:
> hi jeff,
> 
> welcome to nova :)

Good to be here!

> which version of scons are you running? i just figured out that --help
> is broken in the latest scons snapshot 0.97.0d2007something ... it used
> to work in 0.97, though ... 
> maybe i should start bundling scons :/

My bad.  I saw the standard config lines running by and cntrl-c'd before 
seeing the help options after the config.  Odd that it has to check for 
libraries before printing out help information though...  I'll have to 
learn a bit of scons one of these days.

>> * Is there a way to specify the path to portaudio in the build?  (Using 
>> Ubuntu Gutsy I don't have the dev files for libportaudiocpp, and it 
>> would have been nice to just put the library into trunk/libs for 
>> compilation, or at least /usr/local...)
> 
> the headers for portaudiocpp are provided by the portaudio19-dev package
> in both debian and ubuntu ...
> i thought about including portaudio into the nova repository, but
> haven't really been looking into it, yet ...

Hmmm, I thought I tried that and it didn't work.  Maybe I spaced out on 
something.  I'll look into it.

>> * In a simple two object patch with an oscillator to a dac I get 
>> dropouts when I move around gui items.  Is everyone getting this?  Any 
>> idea what's going on?  It seems odd given the separation of the engine 
>> and the gui.  Are position changes being sent to the back-end?
> 
> that is odd ... the information of the position is passed to the engine,
> but with a low priority ... the container that is storing the data is
> using a reader-write-lock, but the lock is not acquired by the realtime
> thread ...
> which driver-backend are you using? 

Portaudio -> HW:0....
   (No idea what driver backend this corresponds to.)

When I try running nova after having started jackd I get a core dump. 
Haven't run with the debugger because I don't know how to do that in 
python yet.

Why does gui information need to travel back to the engine?  It seems 
like the beauty of the separation is that you can just do synthesis in 
the back-end while letting python deal with gui stuff.  Actually, I was 
thinking in the long run the best would be to have scriptable widgets, 
something akin to scripting html with javascript.  My guess is that the 
goal here is to eventually use patches to program gui items?  Seems like 
an unfortunate coupling, although I admittedly don't have a clean 
solution in mind.  I'll think on this one...  Be great to hear what 
people think about this though.

>> * Shouldn't the title of the project wiki get changed to Nova?
> 
> yes ... it is just that i am not a web designer :)

Hmmm.  It seems like hooking artists and musicians is key for a project 
like this, and many will respect a sense of style...  Is the wiki styled 
out with a template and some CSS?

> the only two gui-objects that are currently available are gui.bang and
> gui.comment (you have to type that into the object box)

Cool, I like that.  Using the "gui." will make it especially easy to 
supply IDE style drop-down hints for object name completion as well.

> you are talking about object palettes like in max, aren't you? i would
> appreciate having such a feature, but i wouldn't start coding it
> myself :)
Added to Mantis.

> context menus for objects for help/properties would be nice to have, but
> don't exist at the moment ... 
Added to Mantis.

>> Can we not have path's that suck like PD? :-)
> 
> what kind of paths?
Search paths, location of installed binaries etc.  PD compiles the path 
to pd-gui, for example, into the pd executable.  Since this kind of 
software is so based on externals, libraries, abstractions, patches, 
etc., intelligently dealing with the search path seems important.  For 
example, automatically adding directories to the search path if any file 
has been loaded out of that directory, allowing for a .novarc and/or 
environment variables.  Allowing for run-time additions to the search 
path...  Mostly basic stuff.  This is all done easily in python, but I 
was recently frustrated with PD so I wanted to mention it.

> it would be great if you would be interested in helping with that ...
> pyqt should provide the QUndoStack:
> http://www.riverbankcomputing.com/Docs/PyQt4/html/qundostack.html

Ok.

> only very little of the object handling is actually implemented with qt
> actions, though ... basically only the menus and shortcuts
> 
> if you say, all undoable operations should be implemented as actions, it
> would make sense to make a list of the operations that should be
> undoable and convert them to actions ...

I'm new to Qt so I'll have to read some more, do some experimenting and 
get back to the list on this.

>> I'll be using git-svn to manage my source for the project.  Any other 
>> git users out there?  Tim, you ever think about moving to a distributed 
>> versioning system?
> 
> i just switched to git last weekend :)
> http://article.gmane.org/gmane.comp.audio.nova.devel/188

Nova gets better every day!  I'm compiling out of a newly cloned git 
repo as I type.

> i would be really happy if you find some time to join the development of
> nova ... a few people have offered to work on features of the user
> interface now (mischan - numberbox, niklas - message box, phil - gui in
> general) ...
> personally i started to work on the dsp engine again ... (dsp context
> cleanups, resampling, ffts, multicore utilization) ...

Cool.  This is already feeling like a better fit.

-Jeff
_______________________________________________
nova-dev mailing list
[email protected]
http://klingt.org/cgi-bin/mailman/listinfo/nova-dev
http://tim.klingt.org/nova

Reply via email to