Add PlutusTx.Optimize.Inline.inline for callsite inlining

[zliu41]

This works like GHC.Magic.inline, and can be used in the form of inline f or inline (f args).

[Unisay]

This works like GHC.Magic.inline, and can be used in the form of inline f or inline (f args).

Thanks for adding this! To help me understand the motivation behind this change, could you describe the issue it's designed to resolve? A brief explanation of the problem would be really helpful.

[zliu41]

could you describe the issue it's designed to resolve?

See the tests - the more you inline, the cheaper.

[Unisay]

See the tests - the more you inline, the cheaper.

Its not clear to me why GHC.inline is not enough? Why having our own magic is better?

[ana-pantilie]

@Unisay I think the idea is that we want inline to have the {-# OPAQUE inline #-} pragma, in order to hook into it at the plugin level. We need to define it ourselves if we want that, since we can't modify the source in GHC.inline.

[zliu41]

@Unisay GHC's inline may work, and probably does work most of the time, but there's no guarantee that GHC won't transform it away before it gets to the plugin.

[effectfully]

Great stuff.