Fix Tricks.jl to work for closures

[NHDaly]

Before this PR, the static functions wouldn't work for closures, because their type doesn't have a singleton type.instance defined (since there can be more than one instance!):

julia> make_add_n(n) = x->x+n
make_add_n (generic function with 1 method)

julia> func = make_add_n(2)
#3 (generic function with 1 method)

julia> typeof(func)
var"#3#4"{Int64}

julia> typeof(func).instance
ERROR: UndefRefError: access to undefined reference
Stacktrace:
 [1] getproperty(x::Type, f::Symbol)
   @ Base ./Base.jl:28
 [2] top-level scope
   @ REPL[13]:1

julia> hasmethod(func, Tuple{Int})
true

julia> static_hasmethod(func, Tuple{Int})
ERROR: UndefRefError: access to undefined reference
Stacktrace:
 [1] getproperty(x::Type, f::Symbol)
   @ Base ./Base.jl:28
 [2] #s1#1
   @ ~/.julia/dev/Tricks/src/Tricks.jl:20 [inlined]
 [3] var"#s1#1"(T::Any, ::Any, f::Any, t::Any)
   @ Tricks ./none:0
 [4] (::Core.GeneratedFunctionStub)(::Any, ::Vararg{Any, N} where N)
   @ Core ./boot.jl:571
 [5] top-level scope
   @ REPL[16]:1

After this PR, it works correctly. 😊

I had to copy some functionality over from Base, because they don't provide easy utility functions that work on function-types, only on function instances, so we had to build that ourselves (since we have only function-types, thanks to the generated function).

[MasonProtter]

Looks good to me! Just a couple legibility suggestions based on my personal taste. Feel free to ignore them though if you disagree.

[NHDaly]

Thanks @MasonProtter - have another look if you can? :) i don't have the power to re-request a review, unfortunately.

[MasonProtter]

The implementation is good so far as I can tell.

The only thing I can think of is maybe also add a test with a closure that captures a Box'd value or a Ref? I know it shouldn't matter, but sometimes weird things can be different when a value isn't inline.

We should also think about just dropping support for versions less than v"1.3". IMO it is dangerous and incorrect that these functions used to not support closures.

[NHDaly]

dangerous

well, thankfully they just threw an error, so i don't think it's dangerous?

but yeah, i was surprised to see support for julia < 1.3 as well.

[NHDaly]

The only thing I can think of is maybe also add a test with a closure that captures a Box'd value or a Ref? I know it shouldn't matter, but sometimes weird things can be different when a value isn't inline.

can you suggest a test? It'd be easier than me trying to guess at what you have in mind i think! ❤️

[oxinabox]

Urg, looks like this repo doesn't have GHA setup.
I am loath to merge this if I can't check CI.
Can I leave this open til I find time to add CI?

[oxinabox]

I have given you merge rights.
And I have added CI to master.
Rebase and merge when CI passes?

[NHDaly]

Awesome - thanks!! :)

[codecov-commenter]

Codecov Report

Merging #30 (88f65df) into master (7349d24) will decrease coverage by 1.56%.
The diff coverage is 95.83%.

@@             Coverage Diff             @@
##            master      #30      +/-   ##
===========================================
- Coverage   100.00%   98.43%   -1.57%     
===========================================
  Files            1        1              
  Lines           45       64      +19     
===========================================
+ Hits            45       63      +18     
- Misses           0        1       +1     

Impacted Files	Coverage Δ
src/Tricks.jl	98.43% <95.83%> (-1.57%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7349d24...88f65df. Read the comment docs.

[NHDaly]

Cool, i'm glad you added CI, since this wasn't working on julia 1.3 (or any versions < v1.6).

I've fixed it now, but it's even uglier 🙈
I do wonder if there's some higher-level public interfaces that I missed here, that we should have been calling instead?

If you don't mind, I'll leave this PR open for another round of reviews from you both to see if maybe you can see some cleaner APIs to call that I missed? 🙏 thanks!

[coveralls]

Pull Request Test Coverage Report for Build 1174594970

• 0 of 16 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-2.3%) to 97.674%

---

Totals
Change from base Build 1173795341:	-2.3%
Covered Lines:	42
Relevant Lines:	43

---

💛 - Coveralls

[coveralls]

Pull Request Test Coverage Report for Build 1174594970

• 16 of 16 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-2.3%) to 97.674%

---

Totals
Change from base Build 1173795341:	-2.3%
Covered Lines:	42
Relevant Lines:	43

---

💛 - Coveralls

[coveralls]

Pull Request Test Coverage Report for Build 1275872199

• 18 of 18 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 100.0%

---

Totals
Change from base Build 1222338592:	0.0%
Covered Lines:	51
Relevant Lines:	51

---

💛 - Coveralls

[oxinabox]

let's do it, can always clean up later

[MasonProtter]

LGTM

[NHDaly]

Alright - sorry for my delay. :) Merging now!