Should we avoid using non-ascii, i.e. utf-8 characters in our coding style?

[TeroFrondelius]

@ahojukka5 issues are easy to close. I think it's better to have one topic per issue, it's easier to keep track of them.

Anyway this discussion was started in #3 and this relates also to #5.

I don't have a clear opinion. Both ways have their advantages. For human readability full letter set is better, it makes the reading of the code comparable to the scientific paper, then again avoiding strange characters would probably make debugging much easier.

[ahojukka5]

http://julia.readthedocs.org/en/latest/manual/unicode-input/

There's a self-explaining table how to use utf8 characters and I think in future we will see a lot of these. Of course we could see a lot of difficulties when debugging, 𝘩 != h (U+1D629 != "h"), if we take advantage of full unicode character table. Firefox is not even correctly showing all the characters (today). Notebook rendering engine seems to be broken and drawing red rectangles for every non-ascii character. And this Coverage.jl.

In the other hand, when using e.g. creek symbols like α, β, γ, there's no risk of confusion. At least when not using 𝛂, 𝛃, or 𝛄. While they have a clear mathematical sense they are hard to type without good support of editor / notebook, meaning that user has to copy-paste variables from some sort of unicode table.

Maybe we should consider something like we don't use them in actual code but in notebooks they can be used for clarity?

[ovainola]

@ahojukka5 I think that is the safest option to do. Althought it diminishes readablity in the code, reliability increases. At least with some editors code seems to break down if there's added greek symbols since they can't render them.

Offtopic: I added one greek symbol in the shape_functions.jl to test coverage.jl. This time it didn't break down but scipped the file containing greek symbol. I don't know did it do it because there were no tests available for that for file or because it had that symbol...

[TeroFrondelius]

Let's vote to close this issue. I vote NO to non-ascii. @ahojukka5 and @ovainola your votes please?

Whoever will close this, please remember to add this information to the https://github.com/ovainola/JuliaFEM/blob/master/CONTRIBUTING.md

[ahojukka5]

NO, we should not use them in actual program code.

[ovainola]

I say NO to non-ascii code.

[ahojukka5]

No utf-8 characters to program code. Updated contributing.md.

[TeroFrondelius]

Here is the test I added to keep control. c752ea1