added ⋆ as unary operator

[chakravala]

as disccussed in star is not an operator

[MasonProtter]

Can it be a unary-and-binary-op instead? The Moyal Star product is a legit use of this operator. I don't feel strongly about this though, so if there's a good argument to not bloat the list of unary-and-binary-ops feel free to ignore me.

[chakravala]

@MasonProtter good point, I just committed the change

[JeffBezanson]

For this to work as an infix operator it also needs to be added to a precedence level at the top of the file.

[chakravala]

the prec-times precedence is probably the most appropriate, given that Moyal product is a "product", so that's what I have committed for now

[chakravala]

any update on whether ⋆ could still be merged before v1.2 is frozen?

[StefanKarpinski]

So you want the unary operator version of this to be prefix not postfix? Isn't the traditional Kleene star notation postfix?

[improbable-22]

This PR seems to be about \star which was already a binary operator, and is now listed twice in prec-times.

But the discourse thread was mostly about \bigstar ★. Which would also be nice to have.

[chakravala]

Yes, prefix (not postfix) is used for the Hodge star operation that I implemented in Grassmann.jl

julia> using Grassmann

julia> @basis ℝ^3
(⟨+++⟩, v, v₁, v₂, v₃, v₁₂, v₁₃, v₂₃, v₁₂₃)

julia> ⋆(v1), ⋆(v2), ⋆(v3)
(v₂₃, -1v₁₃, v₁₂)

@stevengj recommended to use \star instead of \bigstar in the discourse

[MasonProtter]

@improbable22 is right, \star is already a binary operator. My bad.

julia> ⋆(x,y) = (x+y)/(x - y)
⋆ (generic function with 1 method)

julia> 1 ⋆ 2
-3.0

[chakravala]

So it should not be added to unary-and-binary-ops and prec-times; while unary-ops is sufficient?

[JeffBezanson]

Oh, sorry I misunderstood --- the linked discourse post is about adding \bigstar, so I assumed this was adding that new operator.

[chakravala]

alright, I have reverted the last 2 commits then

[stevengj]

I think both ⋆ and .⋆ still need to be added to unary-and-binary-ops (in addition to unary-ops).

[chakravala]

are you sure? because this does work for me without it

julia> ⋆(x) = x
⋆ (generic function with 1 method)

julia> ⋆8
8

julia> ⋆(x,y) = x+y
⋆ (generic function with 2 methods)

julia> 7⋆8
15

I also get some compile errors when I have unary-and-binary-ops also, so I think it's not needed

Sysimage built. Summary:
Total ───────  83.465780 seconds 
Base: ───────  27.608859 seconds 33.0781%
Stdlibs: ────  55.854563 seconds 66.9191%
    JULIA usr/lib/julia/sys-o.a
Internal error: encountered unexpected error in runtime:
UndefVarError(var=:li)
rec_backtrace at /home/flow/julia-char/julia/src/stackwalk.c:94
jl_throw at /home/flow/julia-char/julia/src/task.c:210
jl_undefined_var_error at /home/flow/julia-char/julia/src/rtutils.c:130
jl_get_binding_or_error at /home/flow/julia-char/julia/src/module.c:290
_uncompressed_ast at ./reflection.jl:906
typeinf_ext at ./compiler/typeinfer.jl:560
typeinf_ext at ./compiler/typeinfer.jl:599
unknown function (ip: 0x7fc1eb33e417)
jl_apply_generic at /home/flow/julia-char/julia/src/gf.c:2191
jl_apply at /home/flow/julia-char/julia/src/julia.h:1604 [inlined]
jl_type_infer at /home/flow/julia-char/julia/src/gf.c:207
jl_compile_method_internal at /home/flow/julia-char/julia/src/gf.c:1773
jl_apply_generic at /home/flow/julia-char/julia/src/gf.c:2196
jl_apply at /home/flow/julia-char/julia/src/julia.h:1604 [inlined]
jl_finish_task at /home/flow/julia-char/julia/src/task.c:167
jl_threadfun at /home/flow/julia-char/julia/src/partr.c:217
start_thread at /usr/lib/libpthread.so.0 (unknown line)
clone at /usr/lib/libc.so.6 (unknown line)
Generating precompile statements...

this error doesn't happen when unary-and-binary-ops is not used

this PR should be okay as-is right now

[JeffBezanson]

I also get some compile errors when I have unary-and-binary-ops also, so I think it's not needed

That error is definitely not related to this (fix: #31613).

Try [1 ⋆2] and [1 ⋆ 2]; I think those cases are affected by it.

[chakravala]

Alright, here is before

julia> [1 ⋆2], [1 ⋆ 2]
([3], [3])

and after

julia> [1 ⋆2], [1 ⋆ 2]
([1 2], [3])

you are right, this case needs to be resolved, so I have added it to unary-and-binary-ops again

[chakravala]

any more comments? is this PR approved?

[chakravala]

any more comments on the star operation? hoping to have this merged for v1.2