I think this is the general idea of what everyone wants to support. But the way is actually takes shape is going to depend on whoever actually does the work. A great example of this is the PDDP (Pure Data Documentation Project). We had lots of design meetings and then no one implemented the ideas. Then Jonathan picked up from that what was interesting to him and made the whole meta help system, the search plugin, etc.
The lesson there for me is that big design discussions only work if the people involved them are willing to do the work to implement them. Instead, I think for a more decentralized community like this one, we only should nail down the key parts that everyone must use, then leave other decisions to those who are implementing those parts.
So that means I'm happy to help people write there own GUI, and I'll definitely be involved in the work of making it possible with Pd.
.hc
On 01/21/2013 01:05 PM, Leandro da Mota Damasceno wrote:
That sounded like a Lego approach. :)
So the way I see it the GUI development should be in the most seemless way for the user, right?
And we also have the problem between people who prefer a simple, leaner GUI approach (the classic PD, for instance) against people who prefer a more sofisticated, and sexy GUI. And I believe both groups would also like some more knobs and stuff...
so basically, we should at least have two options of gui: simple (classic) or sophisticated (sexy). But it would be cool to make it open enough to anyone develop their own or come up with new and customized ones. that would make PD way cooler than Max/MSP or anything else. So for that to work (and now I must admit I really don't know the architecture behind this part of PD, so maybe it is already this way), the comunication between the GUI and the rest of PD should be kept simple, fast and modulated, working with the leanest possible API. I also think this is a good approach considering that most of these toolkits will stop getting support way before PD ceases to exist. I have also thought about the possibility of skins, but then loading a bunch of bitmaps would not help in terms of performance...
At the same time we pick a toolkit and focus on that one first. So we should think of at least two teems, right? One at the GUI end and the other at the core PD end...
What do you guys think?
On Mon, Jan 21, 2013 at 2:02 PM, Hans-Christoph Steiner hans@at.or.atwrote:
On 01/21/2013 12:54 AM, Jonathan Wilkes wrote:
----- Original Message -----
From: Billy Stiltner billy.stiltner@gmail.com To: IOhannes zmölnig zmoelnig@iem.at Cc: pd-list@iem.at Sent: Sunday, January 20, 2013 10:04 PM Subject: Re: [PD] GUI toolkits and custom GUIs WAS: Integra Live 1.5
released
haha , last month i tried to install juce to see about making an alternate graphics front end to my patches. there was some weirdness in the way you compile it then run the introjucer or somethin to update it then after the update something didn't quite work right. then there are all the old projects that use the old steinberg vst sdk which you cant get from steinberg anymore so all that is just awful. i think that there should be a really nice updated version of juce either available now or in the near future. its a tossup between fltk, qt , opengl ,juce, and processing. i just want to be able to add my waveform data filenames to the presets with a fileopen dialog without using an external, string parsing like .scl files that have 100.00 or 3/2, and polyphonic patchcords would be nice.
What about the -guicmd "cmd..." flag? Could one write a pd-gui.html that lives at localhost:1234, and have it talk to pd at its port on
localhost?
Then you could just write the interface with html5 canvas, svg, javascript, or whatever.
-Jonathan
That sounds feasible to me.
.hc
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list