Use Macros to define new variables

[lllx125]

Now setunits works as a macro that accepts a unit system and reassigns constants exported from AtomicAndPhysicalConstants with value in the proper unit.

Example:
@setunits MKS

This is because @setunits is called in init(), so I can't define constants in a function. However, this makes the constants type unstable.

The code should work properly if we ignore typing.

[DavidSagan]

Fixes problem with AcceleratorLattice.

[DavidSagan]

I am getting the error for atoms:

julia> Species("He")
ERROR: type AtomicSpecies has no field mass_in_amu

[mattsignorelli]

We can’t ignore the types or reassign constants, that is very dangerous and does not work. Instead @setunits should be called in the global scope of the package using AAPC, not in init. Then this will work just fine.Sent from my iPhoneOn Nov 19, 2024, at 10:16 PM, David Sagan ***@***.***> wrote: I am getting the error for atoms: julia> Species("He") ERROR: type AtomicSpecies has no field mass_in_amu —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because your review was requested.Message ID: ***@***.***>

[DavidSagan]

chargeof is broken:

julia> f = Species("electron")
Species("electron", -1 e, 0.51099895069 MeV c⁻², 0.5 ħ, -9.2847646917e-24 J T⁻¹, 0)

julia> chargeof(f)
ERROR: DimensionError: e and MeV c⁻² are not dimensionally compatible.
Stacktrace:
 [1] #s103#141
   @ ~/.julia/packages/Unitful/GYzMo/src/conversion.jl:7 [inlined]
 [2] var"#s103#141"(::Any, s::Any, t::Any)
   @ Unitful ./none:0
 [3] (::Core.GeneratedFunctionStub)(::UInt64, ::LineNumberNode, ::Any, ::Vararg{Any})
   @ Core ./boot.jl:707
 [4] uconvert(a::Unitful.FreeUnits{(e,), 𝐈 𝐓, nothing}, x::Quantity{Float64, 𝐌, Unitful.FreeUnits{(c⁻², MeV), 𝐌, nothing}})
   @ Unitful ~/.julia/packages/Unitful/GYzMo/src/conversion.jl:72
 [5] (::AcceleratorLattice.var"#6#8"{Unitful.FreeUnits{(e,), 𝐈 𝐓, nothing}})(species::Species)
   @ AcceleratorLattice ~/.julia/dev/AtomicAndPhysicalConstants/src/set_units.jl:140
 [6] top-level scope
   @ REPL[6]:1

[lllx125]

Allow macro to be set with keyword parameters

[DavidSagan]

With the latest code I get:

julia> using AtomicAndPhysicalConstants
Precompiling AtomicAndPhysicalConstants...
Info Given AtomicAndPhysicalConstants was explicitly requested, output will be shown live 
┌ Warning: Symbol ħ was found in multiple registered unit modules.
│ We will use the one from AtomicAndPhysicalConstants.NewUnits.
└ @ Unitful ~/.julia/packages/Unitful/GYzMo/src/user.jl:715
┌ Warning: Symbol ħ was found in multiple registered unit modules.
│ We will use the one from AtomicAndPhysicalConstants.NewUnits.
└ @ Unitful ~/.julia/packages/Unitful/GYzMo/src/user.jl:715
┌ Warning: Symbol ħ was found in multiple registered unit modules.
│ We will use the one from AtomicAndPhysicalConstants.NewUnits.
└ @ Unitful ~/.julia/packages/Unitful/GYzMo/src/user.jl:715
... And lots more lines like this...

julia> f = Species("electron")
Species("electron", -1 e, 0.51099895069 MeV c⁻², 0.5 ħ, -9.2847646917e-24 J T⁻¹, 0)

julia> chargeof(f)
ERROR: UndefVarError: `chargeof` not defined in `Main`
Suggestion: check for spelling errors or missing imports.
Stacktrace:
 [1] top-level scope
   @ REPL[3]:1

Also if a "null" species can be defined that would be helpful.

[mattsignorelli]

@DavidSagan Don't you need to call @AAPCdef first?

I'm not sure about the hbar thing though

[DavidSagan]

@mattsignorelli Yes I had forgot that. The package should be configured to throw an error message in this case that tells the user they need to use @AAPCdef.

[DavidSagan]

Also still getting the error:

julia> Species("He")
ERROR: type AtomicSpecies has no field mass_in_amu

[mattsignorelli]

@mattsignorelli Yes I had forgot that. The package should be configured to throw an error message in this case that tells the user they need to use @AAPCdef.

That wouldn't work because massof and chargeof must not exist anywhere before calling the macro.

[DavidSagan]

@mattsignorelli Can dummy functions be defined initially with the use statement to throw an error?

[DavidSagan]

One more point: The default for @AAPCdef should be that C_LIGHT, chargeof, etc be numbers and not have units.

[mattsignorelli]

@mattsignorelli Can dummy functions be defined initially with the use statement to throw an error?

This is dangerous because then every package using AAPC will need to do type piracy on these functions

[DavidSagan]

One more problem: @AAPCdef works outside of AcceleratorLattice but when I try using AcceleratorLattice (with local changes to AL) I get:

julia> using AcceleratorLattice
Precompiling AcceleratorLattice...
Info Given AcceleratorLattice was explicitly requested, output will be shown live 
ERROR: LoadError: syntax: `global const` declaration not allowed inside function around /Users/dcs16/.julia/dev/AtomicAndPhysicalConstants/src/set_units.jl:140
Stacktrace:
 [1] top-level scope
   @ ~/.julia/dev/AcceleratorLattice/src/AcceleratorLattice.jl:26
 [2] include
   @ ./Base.jl:557 [inlined]

The code in AL that triggers this is:

  function __init__()
    @AAPCdef;
  end

[DavidSagan]

@mattsignorelli Since AAPC is controlling things, I would not call this type piracy. In this case, it is definitely not dangerous. From the Julia manual:

Sometimes, coupled packages may engage in type piracy to separate features from definitions, especially when the packages were designed by collaborating authors, and when the definitions are reusable.

[mattsignorelli]

function init()
@AAPCdef;
end

This is the problem, @APCdef should not be called in __init__ but in the global scope of accelerator lattice.

However @lllx125 the macro should just define const, not global const

[DavidSagan]

I removed the __init__ from AcceleratorLattice and the global from APC and the global const declaration problem went away.

[DavidSagan]

@lllx125 You need to change export @AAPCdef to export @APCdef