Sarah Angliss

A request for some help from anyone ofay with bespoke object creation - and the alternatives:

I've been using Max for well over a decade and many moons ago, created some bespoke objects (for a research project) using the Max-SDK that was around then. I expect that was with Max 4 as these were made so long ago. Since then, I've not been involved in any projects that required my own exernals. I've managed with existing objects, augmented with the odd bit of Java / Javascript to get whatever I need.

I now need to create some objects again, this time to work with Max 8 (or 7) - but I'm checking in again at the SDK and can see there are several options available. The old-style SDK, something called min-dev and other options. I'm also seeing a plethora of other coding options available, ones that do not require objects. Some of these look familiar, some look new.

If anyone out there can list all the current coding options in one place, with a simple - 'this uses language x- good if you want to do y' - I'd be really grateful. It would also be good to know which coding setups are tried and tested. Although I'm into trying new stuff, I'm wary of spending a lot of time wrestling with something untested that has minimal documentation - I need to know it's my code that's up the creek, not the setup.

I'm ofay with Java, C and so on - but confused about which way to jump. In my case, I want to do some off-line bespoke data processing, with big matrices, plus some real-time processing of audio signals. Hence going down the DIY object route (but not wedded to that if there's another method). The stuff I'm making is beyond anything I'd want to make via patching existing objects.

Hello 

A request for some help from anyone ofay with bespoke object creation - and the alternatives:

I've been using Max for well over a decade and many moons ago, created some bespoke objects (for a research project) using the Max-SDK that was around then. I expect that was with Max 4 as these were made so long ago. Since then, I've not been involved in any projects that required my own exernals. I've managed with existing objects, augmented with the odd bit of Java / Javascript to get whatever I need.

I now need to create some objects again, this time to work with Max 8 (or 7) - but I'm checking in again at the SDK and can see there are several options available. The old-style SDK, something called min-dev and other options. I'm also seeing a plethora of other coding options available, ones that do not require objects. Some of these look familiar, some look new.

If anyone out there can list all the current coding options in one place, with a simple - 'this uses language x- good if you want to do y' - I'd be really grateful. It would also be good to know which coding setups are tried and tested. Although I'm into trying new stuff, I'm wary of spending a lot of time wrestling with something untested that has minimal documentation - I need to know it's my code that's up the creek, not the setup.

I'm ofay with Java, C and so on - but confused about which way to jump. In my case, I want to do some off-line bespoke data processing, with big matrices, plus some real-time processing of audio signals. Hence going down the DIY object route (but not wedded to that if there's another method). The stuff I'm making is beyond anything I'd want to make via patching existing objects.

Thanks!

Sarah 
www.sarahangliss.com



Hello 

A request for some help from anyone ofay with bespoke object creation - and the alternatives:

I've been using Max for well over a d

max-sdk-for-bespoke-objects-and-alternatives-any-summary-current-options

Hi Sarah, it sounds like you can probably stick with the regular C-based SDK:

https://cycling74.com/sdk/max-sdk-8.0.3/html/index.html

if you've already worked with it(the only difference between now and back when you used it for Max4 is that it gives more access to the increased functionalities added to Max after Max4(like "internal timing mechanism", "dict", etc....), and that it is now using 64-bit). If you're already familiar with C, none of this new stuff added to the SDK will matter much as it's just a matter of reading about those specific features in the SDK once again and reworking code around some newer styles.

The min-devkit is exactly like the SDK, too, except you can use C++(which is much like C except you can program in a more object-oriented style), yet the min-devkit is still in beta and is not as completely documented as the normal SDK(basically, if you really want to program in C++, the min-devkit makes it easy but you'll want to study the normal SDK first so that you're familiar with what's available in documentation).

Other than that, you could also use gen~(there are many things you can do using 'codebox' within gen~ which involves a stripped-down language much like C and still runs very fast). I've still found some things easier and faster in regular C than in gen~(these are mainly things like telling certain functions to be 'inline'... very small stuff that makes some difference in processing speed but for most people wouldn't matter much).

If you want to process data in matrices plus run things at audio rate, it might be easiest to code a hybrid jitter/audio object in C using the regular old SDK. Otherwise, you could try splitting matrix-processing into jit.gen, and audio into regular gen~ and get them to work together somehow(depending on what you want to do, this might give the most immediate gratification as you'll be able to modularize the matrix processing away from the audio-processing, and test them all separately, plus coding in jit.gen/gen~ is easier and faster than using the C or the C++(min-devkit) SDKs), but if you want something more specialized whereby matrix processing can pass data at audio-rate to the audio part of your algorithm directly, then i'd recommend just going straight for the power and speed of regular C(or C++ in min-devkit if you're feeling familiar enough with the normal C-based SDK and you're much more comfortable with C++ already).

I've probably written too much already, but just to summarize: you could at least try brainstorming your ideas in jit.gen and gen~ to see if it's simple enough to implement there, and if you need more power, speed, and specificity, then move in to the C-based SDK(only use min-devkit if you're extremely comfortable with the tools available in the regular C-based SDK already, and you enjoy the object-oriented programming of C++ more than anything else... in terms of speed and efficiency there's no difference between using C and C++ in this context).

Hi Sarah, it sounds like you can probably stick with the regular C-based SDK:
https://cycling74.com/sdk/max-sdk-8.0.3/html/index.html 
if you've already worked with it(the only difference between now and back when you used it for Max4 is that it gives more access to the increased functionalities added to Max after Max4(like "internal timing mechanism", "dict", etc....), and that it is now using 64-bit). If you're already familiar with C, none of this new stuff added to the SDK will matter much as it's just a matter of reading about those specific features in the SDK once again and reworking code around some newer styles.
The min-devkit is exactly like the SDK, too, except you can use C++(which is much like C except you can program in a more object-oriented style), yet the min-devkit is still in beta and is not as completely documented as the normal SDK(basically, if you really want to program in C++, the min-devkit makes it easy but you'll want to study the normal SDK first so that you're familiar with what's available in documentation).

Other than that, you could also use gen~(there are many things you can do using 'codebox' within gen~ which involves a stripped-down language much like C and still runs very fast). I've still found some things easier and faster in regular C than in gen~(these are mainly things like telling certain functions to be 'inline'... very small stuff that makes some difference in processing speed but for most people wouldn't matter much).

If you want to process data in matrices plus run things at audio rate, it might be easiest to code a hybrid jitter/audio object in C using the regular old SDK. Otherwise, you could try splitting matrix-processing into jit.gen, and audio into regular gen~ and get them to work together somehow(depending on what you want to do, this might give the most immediate gratification as you'll be able to modularize the matrix processing away from the audio-processing, and test them all separately, plus coding in jit.gen/gen~ is easier and faster than using the C or the C++(min-devkit) SDKs), but if you want something more specialized whereby matrix processing can pass data at audio-rate to the audio part of your algorithm directly, then i'd recommend just going straight for the power and speed of regular C(or C++ in min-devkit if you're feeling familiar enough with the normal C-based SDK and you're much more comfortable with C++ already).

I've probably written too much already, but just to summarize: you could at least try brainstorming your ideas in jit.gen and gen~ to see if it's simple enough to implement there, and if you need more power, speed, and specificity, then move in to the C-based SDK(only use min-devkit if you're extremely comfortable with the tools available in the regular C-based SDK already, and you enjoy the object-oriented programming of C++ more than anything else... in terms of speed and efficiency there's no difference between using C and C++ in this context).

Hope it helps, best of luck 🍻


This is really useful - exactly the info I'm after. I hadn't considered codeblocks in gen~ so will definitely see if that's a good fit, as well as the updated SDK.

Thanks Raja
This is really useful - exactly the info I'm after. I hadn't considered codeblocks in gen~ so will definitely see if that's a good fit, as well as the updated SDK.
Thanks so much for your time!
Sarah


Max-SDK for bespoke objects - and alternatives - any summary current options?