Fix show_after error when sizeof fails to determine DataType size

[Zapiano]

This error is preventing users from creating YAXArrays of Strings} and Unions that contain Strings.

closes #410

[felixcremer]

Thanks for trying to tackle this, but I think, that it would be good to not have a try catch in the show method.

[felixcremer]

If we rethrow the error, we will get a better message but still can't show the cube without an error. So I think that this should not be rethrowing.

[lazarusA]

@felixcremer what's is blocking this?

[felixcremer]

I don't like the use of try catch in this.
Could we use Base.summarysize(first(c)) instead of sizeof(eltype(c)) in the definition of cubesize?
This should give the same result for plain types like Int and Float64 and gives an approximate size for other types like string.

[felixcremer]

The problem is, that we would introduce a getindex call which might be even more problematic than the use of try catch. So I am fine with that or with removing this info and only enabling that for certain eltypes where we know that sizeof is giving the right answer.
We could do that by putting the lines 511 to 513 together into a print_cubesize function which can then dispatch to the eltype of the cube and which can default to nothing to not print this information.

[Zapiano]

cc: @lazarusA

The problem is, that we would introduce a getindex call which might be even more problematic than the use of try catch. So I am fine with that or with removing this info and only enabling that for certain eltypes where we know that sizeof is giving the right answer. We could do that by putting the lines 511 to 513 together into a print_cubesize function which can then dispatch to the eltype of the cube and which can default to nothing to not print this information.

Thanks for the response and sorry for the delay.

I made a change checking for some specific eltypes for which the function sizeof is known to behave well before printing the size of the datacube. Let me know if that approach is better or what you think I need to change.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Project coverage is 68.39%. Comparing base (59d23fc) to head (51d5f82).
Report is 70 commits behind head on master.

Files with missing lines	Patch %	Lines
src/Cubes/Cubes.jl	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #421      +/-   ##
==========================================
+ Coverage   68.30%   68.39%   +0.08%     
==========================================
  Files          12       12              
  Lines        1773     1778       +5     
==========================================
+ Hits         1211     1216       +5     
  Misses        562      562              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.