Compiled mode doesn't support second outlet of [inlet~]

[brianhouse]

Using the second control output of an inlet~ results in a compile error. Perhaps an issue or an intentional implementation thing in hvcc, but there's no indication in plugdata that this is the case, and if it's several abstractions down it's tricky to find.

[dromer]

I've actually never seen anyone use this, but indeed both the control inlet and outlet are not supported in Heavy on the inlet~ object. This should certainly be documented better, so I'll have a look at this.

I don't know how we could detect such things in plugdata compiled-mode though.
There are also other things that won't work when compiling, like some specific object arguments or messages, that would be very difficult to detect and give the user a warning about.

[brianhouse]

I'd say that having control messages as well as a signal come in through the first inlet is a very common pattern, especially for people coming from Max

[dromer]

Heavy can't support this control inlet and outlet on the [inlet~] object though. And either way this repository is not the place to request/discuss it ;)

The point is that these kind of highly specific limitations are very hard to detect and warn the user about.
This will require much more advanced methods to analyze patches and specific language to inform the user.

I think it's definitely out of scope for a Plugdata v1.0 release at least, but maybe at some point if someone can figure out both the UX and actual implementation.

I'd definitely like to have both better documentation and better insight for users about the limitations.
Heavy only emulates PD behavior and is not a 1:1 re-implementation.

[brianhouse]

makes total sense. just flagging it.

[dromer]

Hah, so I was actually able to get it to work! Did not think this was possible: https://github.com/Wasted-Audio/hvcc/pull/409/changes

So in the next release we will have both control inlet and control outlet of [inlet~] object working.

The entire test-suite still passes, so I hope that this does not introduce any unintended side-effects.

The main thing is that inlets and outlets are not exactly typical "objects", but this way the parser kind of works around the limitations and adds the required connections anyway.
I've only tested it with a subpatch, but I expect it to work for abstractions as well.

[brianhouse]

🔥