Infix/prefix translation functions

[benbrastmckie]

This issue will follow the development of the infix and prefix functions described in /Design/Strategies.md, providing a place for questions and problems to be raised. All TODOs can be tracked in the TODO.md file. Here are two initial thoughts.

• It would be great if these functions did not depend on which sentence letters have been included so that they could be used in languages with different sentence letters.
• If this is difficult, we could start by assuming the sentence letters A, B, C, D and come back to generalize the functions later.

Feel free to post any related issues or questions that arise here.

[mbuit82]

Sounds good! I'm wondering—as for the Or(x,y) function that's on the strategy doc, why don't we just call it Fusion(x,y)? seems it would be much more transparent, unless there really is a difference between Or and Fusion that is best be reflected later.

[benbrastmckie]

I'm going to create a new issue to preserve this tread for the infix/prefix functions, though of course all of these issues are going to be related.

[mbuit82]

Sounds good—just made a function for translating infix to prefix. There's two issues left to resolve: 1. how bad is it if the incoming representations have forward slashes instead of backslashes? 2. what do we want to do with prefix notation? I put both those questions in the questions.md file also

[benbrastmckie]

If the double slash strategy works, that is probably the simplest, but the zero-slash strategy could also work (or forward slash). Whatever works is great.

We chatted about what the prefix function is for, but I will include some notes in the Strategies.md file since I hope to slowly turn this into documentation.

Also, I'm not sure that I mentioned this, but it matters a lot less what name gets used for the operators in prefix notation since these are only help the semantics decide which recursive clause to use. So Maybe best to have no slashes there. Although tempting to go all the way to And instead of Wedge, this might get confusing since And will also show up to connect Z3 conditions.

As for TODOs, it seem that all that remains is:

• [ ] M clean up function names and unused parts
• [ ] M check that (A \\wedge B) goes to [wedge, A, B] rather than [Wedge, [A, B]]
• [ ] M check that A goes to [A]
  • I'm assuming that this is simplest so that all prefix sentences are lists
  • That way when it comes time to interpret sentences, we just ask if the first list item is an operator.

Seems like we're close to closing our first issue!

[mbuit82]

I've rewritten it (about to push) so that it should in principle be able to accept all of \wedge, /wedge, or wedge. I think this provides more flexibility on the user end and it may actually be easier and more efficient to resolve the translation from those things to something like Wedge after the construction of the infix form. I cleaned up the prefix_infix file, and now A goes to [A]. This has the consequence that (A \wedge B) goes to [wedge, [A], [B]] (ie, args of operators are in 1-elem lists) but this would make a function calling recursively to get the semantics of an infix function easier to write (given we want A to map to [A]), so I think it's actually good. I think that's basically everything on this! (also, how do you make text in the middle of the comment look like code?)

[benbrastmckie]

Awesome looks great! Regarding you question there are two ways. Either you indent four spaces, or wrap with triple quotes.

[benbrastmckie]

Hey was just looking at the prefix function and it looks great. I'm going to close this issue, though we can continue to post here if anything relevant comes up.