chore: using Libdl

[kjappelbaum]

Great work, thanks for the effort, will be useful!

this seems to be needed for Julia > 0.7. At least i needed it to get started with Julia 1.5.

Unfortunately, I run into

julia> net = CrystalNet(crystal);
ERROR: MethodError: no method matching prod(::Type{BigFloat}, ::Array{Float64,1}; init=1.0)
Closest candidates are:
  prod(::Any, ::AbstractArray; dims) at reducedim.jl:723 got unsupported keyword argument "init"

I guess (

CrystalNets.jl/src/specialsolver.jl

Line 238 in 2cc25ad

δ = prod(BigFloat, λs; init=one(BigFloat))

) you use the Julia 1.6-dev? Could you maybe update the dependencies?

[Liozou]

You are absolutely right, it should be working for julia v1.5 as I stated in the the Project.TOML file:

CrystalNets.jl/Project.toml

Line 36 in dd5959a

julia = "1.5"

but the init keyword was only added to Julia in JuliaLang/julia#36188, which did not get merged in time for 1.5. I fixed both this init issue and the spurious Libdl dependency you solve in the PR through f3c279a, so it should now run on 1.5. I prefer to remove the dependency altogether from the precompile.jl file rather than add it explicitly because this precompile.jl needs some cleanup anyway.

Overall this package is still quite experimental, which is why it is not registered yet, but please feel free to use it if you want to! And I appreciate the feedback a alot, thank you for this.

[kjappelbaum]

Many thanks for the swift reply!

Ok, I'll try again with my 1.5 installation (ran after a quick test again into the error with init and I mentioned 1.6 as i found it in the 1.6 changelog: https://github.com/JuliaLang/julia/blob/d83d81e5899c5a81d3bc2e6ccd117dca821c9d1f/NEWS.md)
Anyhow, it seems to work for me by replacing 1.5 with 1.6 in

CrystalNets.jl/src/specialsolver.jl

Line 238 in f3c279a

δ = @static if VERSION > v"1.5+"

Many thanks, it's really something I've been looking for! (so just to understand the status of this package: the test cases you have there are expected to pass right now?)

[Liozou]

Apologies, there were two lingering issues that your comment prompted me to check (thank you!), so that should now be fixed. Please let me know if it works!

The tests are not exactly up to date, so I would expect them to either work or stall indefinitely. Otherwise I'll gladly take the error log to see what's going on.

[kjappelbaum]

Great! Thanks, got the executable working and also got the topologies for some of my test cases right!

It's a great work, if I can help in any way let me know!

[Liozou]

Glad I could help!
Please feel free to open issues if you encounter weird bugs or internal errors with CrystalNets, my goal is to make it as usable as possible. And of course, please let me know if it gets any topology wrong, as that should never happen!