#2024: Fix CI issue with empty CXX standard variable

[thearusable]

Closes #2024

[PhilMiller]

The script changes are fine. Why do we need to add the FORCE option?

[PhilMiller]

Also, what's the relationship between this change and the error shown in the linked issue?

[PhilMiller]

I believe Cezary understands, given his quick approval, but it's helpful to write things down so the rest of us can follow later

[thearusable]

@PhilMiller Force option is for overriding the already existing value in CMAKE_CXX_STANDARD, without that CMake's set function will not modify the value. In our case it was incorrectly set to an empty string.
So now even if we set that variable to "" in the script then it will still work properly because it would override it with the default value of 14.
Previously there was also a small issue in the scripts which will pass "" when no value is set. Now they will set a proper default value.

[PhilMiller]

It seems that if the variable is set with nonsense, we shouldn't carry on and try to do something that looks sensible. If it's a likely user mistake, we should detect it and report the error. Otherwise, I don't think we should try to compensate for a script bug that we know about and are fixing.

Is there some subtlety here I'm missing?

[thearusable]

No, you are right. Ok so I'll remove the FORCE option. Scripts will not create that issue again and if user set some gibberish then it would fail as on the pipeline right now.

[codecov]

Codecov Report

Merging #2032 (286b290) into develop (5dff872) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff            @@
##           develop    #2032   +/-   ##
========================================
  Coverage    84.47%   84.47%           
========================================
  Files          731      731           
  Lines        25847    25847           
========================================
  Hits         21834    21834           
  Misses        4013     4013           

[stmcgovern]

Setting the defaults fixes the script issues just fine. Cf #2025 for the build_cpp.sh

[stmcgovern]

I'll close #2025 since this PR includes that and an additional fix.