On Sun, 25 Jan 2004, Frank Barknecht wrote:

Nice idea, and I think, useful in the future.

I assume this was meant as a reply to the list; I hope you don't mind.

I hope, it does not lead to interrupts in the audio chain on larger post()'s?

Yes it does, like pretty much any heavy activity involving a communication between pd and pd-gui.

Today I tried tweaking sys_gui() in a way that makes it go smoother. Somehow I can't get it to work with fdopen()/fwrite() buffering, I can't get it to work with send(,,,MSG_DONTWAIT), and I can't get it to work with O_NONBLOCK, as in all three cases the data gets badly mangled along the way, and I can't figure out why. I basically wanted to make it so messages get buffered by sys_gui(), and it would return even when a message hasn't been sent yet, and would try again later at a further call of sys_gui(...), and the main loop would try sending the rest when it can, by calling sys_gui(""). I then gave up on getting it to work because of the aforementioned issues.

But personally I see another problem, at least on Linux, with the bigger main window. If I use an openpanel/savepanel in my patches, then the Pd main win will lay itself above all other Pd windows after I selected a file. This is really annoying, because normally I don't need to see the Pd main window at all. Of course this isn't your fault, but an even larger main window would make for an even larger window to click away after selecting files. ;)

You know, I had no clue about this, but I thought that someone would find such a problem eventually, which is why it was important to me that this feature be enabled explicitly by "-console", to ease its integration into the main Pd release.

Besides, Pd does all sorts of other fprintf(stderr,...) that I haven't changed yet; so consider it as an "alpha" feature.

________________________________________________________________ Mathieu Bouchard http://artengine.ca/matju