Primitive port and node construction

[mtrberzi]

I would like to submit this pull request for review, knowing that I am not done with it yet. There is a lot of code here, but finishing all of the features I have marked as TODO in the code would result in a PR of several thousand lines, and this is bad enough already.

• Yes, this finally gets rid of the old digital circuit primitive definitions. Those are going to be gone permanently after this, because we can now define port types and node types in the frontend language.
• This also removes all of the old "functionality" of ExpressionGraph, which made horrible assumptions about the structure of the schematic that was being compiled and wouldn't work for anything except digital circuits.
• TypeChecker is completely gone. All typing can be done dynamically through the expression graph, which has all the benefits of a static system while providing dynamic typing only where it is actually used.
• Test coverage is down quite a bit, but I plan to add some test cases to this branch after it is reviewed and hopefully get the coverage even higher than it is now.
• There are many TODOs hanging around in the code. Some of them are easy enough to resolve before this is merged; some of them are large enough to merit opening a separate issue.

[coveralls]

Coverage decreased (-20.45%) when pulling 65efbb136d2a895a680251e14b99d0ee59c7bcef on mtrberzi:primitive-definitions into 3d9b44aacf7930623630fcccd7d50a5be8ae75b9 on manifold-lang:master.

[maxqchen]

Is it possible to get a rough high level view of the structure, something like pseudo UML? Would make it easier to understand the relations between everything.

[mtrberzi]

As something visual that I had laying around, here's the expression graph for the following code:

xIn = primitive port Bool;
xOut = primitive port Bool;
xNot = primitive node (x : xIn) -> (xbar : xOut);

This includes all the major ExpressionVertex types that ExpressionGraphBuilder deals with. Hopefully seeing it work is a little easier to visualize than UML, but let me know if there's something else I can prepare to help you wrap your head around this one (I know that this is some difficult code to review so I'm ready to assist).

[coveralls]

Coverage decreased (-20.43%) when pulling 6f106fb17d1ef12dcec76eca4a77436a2196a72a on mtrberzi:primitive-definitions into 3d9b44aacf7930623630fcccd7d50a5be8ae75b9 on manifold-lang:master.

[mtrberzi]

Oh, by the way, fixes #31.

[mtrberzi]

Also fixes #16.

[mtrberzi]

Also fixes #17.

[coveralls]

Coverage decreased (-20.43%) when pulling ed5d01b27a38e1fd6243137c1fc64a658cf0ba41 on mtrberzi:primitive-definitions into 3d9b44aacf7930623630fcccd7d50a5be8ae75b9 on manifold-lang:master.

[coveralls]

Coverage decreased (-20.36%) when pulling eef2ea4de18aee9928e006bb7baea8427048636d on mtrberzi:primitive-definitions into 3d9b44aacf7930623630fcccd7d50a5be8ae75b9 on manifold-lang:master.

[lucaswoj]

As a high level point, I really hope we can push for removing the concept of an abstract syntax tree and generate a semantic tree directly.

[lucaswoj]

Done my first review. Sorry about the zoo.

[maxqchen]

Logic looks good (AFAIK anyways, still a bit fuzzy on the whole thing), just a few minor nit picky things :bike:.

[mtrberzi]

Fantastic first round of review, everyone! Thanks again. New commits are available.

[coveralls]

Coverage decreased (-20.15%) when pulling 8786e671a3df40aa900adca2059bc1943f013b7b on mtrberzi:primitive-definitions into 3d9b44aacf7930623630fcccd7d50a5be8ae75b9 on manifold-lang:master.