Remove \ from IR.Type.Set but support &

[JaroslavTulach]

There are some SugaredTypeFunctionsTest that deal with support for sets(!?) and produce IR.Type.Set. Some of the operations seem to be working (like T & P or T | P) but some of them aren't. Particularly the T \ P syntax seems to contradict with the "new syntax for lambdas".

Please take a look at SugaredTypeFunctionsTest three failing tests:

• should resolve context ascription
• should resolve equality
• should resolve subtraction

and decide whether such tests are supposed to be fixed or deleted. Thanks for your help.

[wdanilo]

You are right. The syntax T \ P is not working anymore, as it's needed for lambdas. However, I am not sure we will support full boolean algebra on contexts. In particular, I'm not sure if we can support the \ operator in a general sense. Maybe we could, but until we are sure about it and it is correctly designed, I'd propose to just remove the \ boolean operator and use this syntax for lambdas only.

[JaroslavTulach]

just remove the \ boolean operator and use this syntax for lambdas only.

OK. Let's remove \ operator support. Shall we keep all the others &, |, etc. or attempt to remove this whole IR.Type.Set support?

[wdanilo]

Yes, we should keep them. String | Nothing and similar types are super useful and are foundation for many things we are designing with James.

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-09-08):

Progress: - PR: #7769

• review Integer unification: Merge Small_Integer and Big_Integer types #7636 (review)
• trying to understand IR passes that mangle Type.Ascription
• prototype: x : Text is valid syntax, but doesn't check runtime type of the expression #6848 (comment)
• bookclub It should be finished by 2023-09-12.

Next Day: Bugfixing

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-09-11):

Progress: - removing \ supporting & integrated: #7769

• Attempt to fix StackOverflowError in sort called on numpy array. #7677 by Avoid StackOverflow when comparing unknown foreign objects #7780
• benchmarking and integrating: Benchmark shall loop correctly #7781
• review: Workaround crashes during regular execution #7778 (review)
• demo & other meetings It should be finished by 2023-09-12.

Next Day: Bugfixing