He there. I'm beginning to get acquainted with music production under
linux and there's a question that bugs me about pure-data.

Some argue that one strength of music production under linux is the modular
approach given by the jack audio server; it allows completely independent apps to route signals to each other, etc. etc. The major problem with this is a complete lack of cohesion: there is the classical foss paradigm, you have xyz apps, each providing the appoximately the same functions as the other, are coded in perhaps a different languages, have their trade-offs here and there and then A is missing some must-have feature of B and conversely; lots efforts are dissipated, UIs are a chaos, the softwares architectures are redundant with bugs, duplicated semi-polished features, etc.

Then you have pure data, which at first glance, looks like a high-level, integrated audio developement environnement, just the unified framework one would need to build and enhance a perfect music creation environnement. Yet as it looks (forgive the short sighted vision of a newbie) pd has remained a developement-only platform that has yet to yield it's fruit to that other end of the spectrum of the music community which interests are more focused on the architecture of rythms and melodies instead of the software generating them.

So my question is this: is there such environment and if not, why not?

there are a lot of people making end-product music with pure data.

the main stumbling blocks that prevent it from going much further are, in my opinion:

it's difficult to integrate PD into other music production environments. (reaktor can be built as a VST, max/msp has pluggo, etc), and to make complete works in pd alone requires fairly significant knowledge and effort.

it's difficult to integrate other music plugins in pd.

pd's gui is incredibly primitive. these days, for music production on a computer, it's pretty much essential to have some sort of display and edit functionality. i would love to build a basic multitrack DAW in pd, but the gui would be maxed out way too quickly.

the 'why not' i guess comes down to money. if people were being paid dollars or euros to get this stuff done, pd would have been ported out to a better interface long ago, and been fully integrated as a plugin effect. various people have tried to do these things, but the end results were never too amazing, so the projects mostly disappeared.

that all said....if robert moog and raymond scott had been happy just to use their normal instruments, music would not be the same as it is today. it is the people who take the less travelled roads that make all the difference. (to paraphrase some dodgy poem). if you think of pd as a bit of a challenge, liek using a modular synth would have been in the late 70's, for example, then it's indeed a world full of unexplored opportunity. of course it's harder to get things to work, but when you do, you inevitably get something with a lot of your own soul and hard work in it. that usually does come across in the sound.

I will attempt to precise the architecture I'm thinking of to ensure we'll really be talking about the same thing.

Layer 1.

The music production environment GUI. Featuring the usual stuff: multi-track classical/midi partition editor, instruments bank, pre/post processing effects chaining, etc. Most importantly, controls should be displayed and respond in a uniform way, there also should be a high level script-language/gui application for designing the interface of new instruments/effects/etc.

Layer 2 (i.e.: Pure Data)

A Self like programming language, highest level as possible, for quickly designing and prototyping new instruments/effects, to be incorporated into level 1.

Layer 3.

An application framework for extending the capabilities of layer 1 and 2.

Now what I find appealing in this model, is that every significant level of the music creation process would be sufficiently organized to be tempered with while remaining on the level one is addressing (i.e.: Although the possibility remains, a musician interested only in composing melodies and stuff, won't have to cope with the low-end technical details of the instruments he's using; the instrument designer won't have to care about the underlying programming framework, etc, etc.)

Now to address some of you comments:

"it's difficult to integrate PD into other music production environments"

• I don't see the need to integrate PD in other music production environment, as to me, PD shouldn't be a mere plugin but the foundation of a music production env.

"to make complete works in pd alone requires fairly significant knowledge and effort."

• As stated above, PD shouldn't be the place for making a complete musical work, but rather for designing one's instruments.

"pd's gui is incredibly primitive" and
" i would love to build a basic multitrack DAW in pd, but the gui would be maxed out way too quickly."

True, and this is why the layer 1 is really needed.

Now, I certainly don't consider the instrument engineering part of music as to be a waste of time; although it would be a waste of time if those instruments couldn't be used by anyone except their makers - and the same goes for any program. I know these suggestions I'm rambling about are probably not new nor would they probably cause any revolution in linux audio, I only hope it could stand as a direction the ones interested in developing PD a little further might seriously consider.

Cheers to Moog and Scott.

Just a thought on the side,

My personal goal with pd is to design a synthesizer and make it the way I want it. No "save as" development software like synthedit, etc, has proven adequate for what I have in mind.

having said that, once I finish, I'm going to do a barebones build of linux on a flash drive with my synth patch. That way, I can turn any computer with a usb drive into a dedicated synth. Run midi to it and get sound out, the old-fashioned way.

Ideally, I'd like to make it look comprehensible enough with gui abstractions and such that anyone can load it onto a linux flash drive and play with it. For anyone with a spare/old/spouse's/stolen laptop, it will be like having a hardware digital synth. No cpu hit on the system you're sequencing with, and midi in, audio out is pretty much universally compatible.

This isn't really addressing the issue you put forth,
but I do feel like its only a matter of pd developers making an understandable gui and then writing nice picture tutorials on how to make it work for the musicians

Also, in defense of the pd visuals, I find the minimalist vectors ten times more attractive than the vast majority of designs vst developers throw together

"I do feel like its only a matter of pd developers making an understandable gui and then writing nice picture tutorials on how to make it work for the musicians"

Indeed my plea is mainly concerning current PD developers and while the programming of a unified GUI is certainly one of the matters I'm concerned with, "nice picture tutorials" cannot compensate for a unified framework with clearly and well separated layers that would encapsulate the information level you have to cope with while doing your stuff. You could build a synth in C - or assembly for that matter, having a nice C or assembly tutorial under your hand, but I guess that would be overkill, right?

"Also, in defense of the pd visuals, I find the minimalist vectors ten times more attractive than the vast majority of designs vst developers throw together"

Minimalism is in no danger here - and I don't think it should be of any concern for the moment. An aspect I would like to consider though, is what information should be allowed to pass through the GUI.

What i personally like so much about pd is the complete lack of a predefined way of working/thinking. You start the runtime, create a blank patch and there you go. Noone tries to force me into the "multitrack" approach. The choice is completely left open to the user. What suits one kind of musician won't per defintion suit someone else. That's also the reason i believe pd is used very often for things that don't have to do with music per se. Artists from many different disciplines seem to be attracted towards pd (or max if you want) because it's an empty playground waiting for you to be filled with your own toys.

It's true however that the gui lacks a bit when compared to modern applications that use the full potential of modern hardware, but this hasn't bothered me at all so far. On the contrary, since pd isn't bloated with fancy pancy stuff it allows you to focus on the actual idea you're trying to implement instead of detracting you to do all kinds of gui fanciness nobody really wants - although its always a nice challenge to create fine looking gui's with the bare objects provided, and it's also always nice to open patches that are nice looking and want you to play with them ...

Hmm. In the case it wasn't made clear enough, I do not suggest a reduction of PD to a more constrained environment but rather, allow it to be integrated in a richer context that would allow it to reach a broader audience - the only constraint needed for this, is, in a way, a common language between the layers.

Anyways, I don't think I can make the general idea clearer. If any developer is interested in the endeavor, I'd be glad to contribute - note that I'm not deluded enough to think I can lift such a project all by myself as I have other, more pressing concerns for the moment.

Again, my point was to poke to see if there is - and perhaps ignit - some interest in this direction. If there's none, then I'm beating a dead horse, linux audio is just fine and dandy... I'll just boot to where I can find a decent grand piano.