Error: type mismatch on echo

[sdickreuter]

Hi !

I'm a newbie with nim and I was trying out some arraymancer examples to get a feel on the usage, but when I run the example code

import arraymancer

let d = [[1, 2, 3], [4, 5, 6]].toTensor()

echo d
# Tensor of shape 2x3 of type "int" on backend "Cpu"
# |1      2       3|
# |4      5       6|

echo d.rank # 2
echo d.shape # @[2, 3]
echo d.strides # @[3, 1] => Next row is 3 elements away in memory while next column is 1 element away.
echo d.offset # 0

compilation fails:

bla.nim(5, 6) template/generic instantiation of $ from here
../../.nimble/pkgs/arraymancer-#head/tensor/display.nim(25, 27) template/generic instantiation of disp3d from here
../../.nimble/pkgs/arraymancer-#head/tensor/private/p_display.nim(96, 17) Error: type mismatch: got <Tensor[system.int], proc (x: GenericParam): untyped>
but expected one of:
proc map[T, S](s: openArray[T]; op: proc (x: T): S {.closure.}): seq[S]
first type mismatch at position: 1
required type: openarray[T]
but expression 't0' is of type: Tensor[system.int]

expression: map(t0, proc (x: auto): auto = $x)
[Finished in 1.2s with exit code 1]

if I remove echo d it compiles and runs fine.
I'm running latest nim from git (1:0.19.4.r1306.c7c495f08-1) and have the issue on arraymancer 5.0 and head.
Any help is appreciated!
cheers
Simon

[chimez]

Adding type hinting of map will fix it.

[mratsim]

AFAIK this was a regression in Nim, that was detected by @Araq when he added popular packages to Nim regression test in nim-lang/Nim@f4c76bc and nim-lang/Nim@6e1d2df.

I tried on latest devel but it's still failing so maybe the fix is in a branch. It works on stable though.

Thanks @chimez for the PR I'll merge it.

[sdickreuter]

Thank y'all for the quick reaction and fix!