WIP: private patch abstractions [ab]. first approach

Thanks, Guillem!

Quite a usability enhancement to create an [ab foo] and start editing it without having to touch the filesystem!

A few immediate things spring to mind:

Any problem with [ab] taking A_DEFSYM? It's just a nice pattern for the user to be able to create without args. For example, sometimes users will instantiate an object just to right-click "Help."
How to handle saving. It took me awhile to get my changes to show up in subsequent abstractions because I hadn't thought to save the original [ab]. Also, it feels a bit weird to successfully save an [ab] in a patch which itself hasn't been saved. But I don't yet have any suggestions on alternative behavior to address this.
It appears that a new [ab] instance with signal objects inside it isn't added to the DSP graph until you open it and create a new object.

1.Any problem with [ab] taking A_DEFSYM? It's just a nice pattern for the user to be able to create without args. For example, sometimes users will instantiate an object just to right-click "Help."

As far as I know, it is not possible to combine a GIMME arg type with other types (e.j. [...] A_DEFSYMBOL, A_GIMME, A_NULL). Therefore, the GIMME type is the only option to allow the abstraction to take arguments.

2.How to handle saving. It took me awhile to get my changes to show up in subsequent abstractions because I hadn't thought to save the original [ab]. Also, it feels a bit weird to successfully save an [ab] in a patch which itself hasn't been saved. But I don't yet have any suggestions on alternative behavior to address this.

Yes, I don't really like it either, but it is the most straighforward way I have come up with so far.

3.1. It appears that a new [ab] instance with signal objects inside it isn't added to the DSP graph until you open it and create a new object.

I'll take a look.

added 1 commit

7b6edd1b - some clean up. getbytes + memcpy + freebytes -> resizebytes

Compare with previous version

Therefore, the GIMME type is the only option to allow the abstraction to take arguments.

Ah, that's correct.

I think it's still worth special casing this in ab_new. If argc = 0, choose a default symbol to bind to. I'm not sure if &s_ would cause problems when saving. If so, there are workarounds for that.