DArray macro fails on 1.5

[raminammour]

Hello,

on the official binary:

julia> using DistributedArrays
[ Info: Precompiling DistributedArrays [aaf54ef3-cdf8-58ed-94cc-d582ad619b94]

julia> @DArray [1 for i in 1:2]
ERROR: syntax: malformed expression
Stacktrace:
 [1] top-level scope at REPL[3]:1

julia> versioninfo()
Julia Version 1.5.0
Commit 96786e22cc (2020-08-01 23:44 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin18.7.0)
  CPU: Intel(R) Core(TM) i9-9880H CPU @ 2.30GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-9.0.1 (ORCJIT, skylake)
Environment:
  JULIA_EDITOR = /Applications/Sublime\ Text.app/Contents/SharedSupport/bin/subl

Is this easy to fix? The error is not very helpful...

Cheers!

[bjarthur]

i can confirm. even on master. the example straight out of the docs fails too, with the same error message. @DArray [i+j for i = 1:5, j = 1:5]

[andreasnoack]

I'm not sure what is going on here. If I print the final expression from the macro then I can eval that without problems but the macro fails. Let's ask @JeffBezanson if he knows what is going on here.

[raminammour]

It is the assertion ::Tuple{Vararg{UnitRange{Int}}} that it does not seem to like.

This works:

@eval DistributedArrays macro mDArray(ex0::Expr)
           if ex0.head !== :comprehension
               throw(ArgumentError("invalid @DArray syntax"))
           end
           ex = ex0.args[1]
           if ex.head !== :generator
               throw(ArgumentError("invalid @DArray syntax"))
           end
           ex.args[1] = esc(ex.args[1])
           ndim = length(ex.args) - 1
           ranges = map(r->esc(r.args[2]), ex.args[2:end])
           for d = 1:ndim
               var = ex.args[d+1].args[1]
               ex.args[d+1] = :( $(esc(var)) = ($(ranges[d]))[I[$d]] )
           end
            
           return :( DArray((I)->($ex0),
tuple($(map(r->:(length($r)), ranges)...))) )
       end

DistributedArrays.@mDArray [1 for i in 1:2]
2-element DArray{Int64,1,Array{Int64,1}}:
 1
 1

Whether the assertion is really needed, or how to meta-program an assertion is beyond me at this point ;)

Cheers!

[raminammour]

This appears to be a julia issue, filed.

[raminammour]

Update: the issue was solved on the julia side, and should make it into 1.5.2, so feel free to close the issue and the related PR if you see fit. Unless you want a version that works for 1.5.0/1.5.1.

[raminammour]

Since 1.5.2 was released, containing the fix on the Julia side, I am closing this issue and the workaround PR. Cheers!