Tests .ab rules

[Mte90]

So we have more then 100~ tests in pure Amber.

My idea is that we define some rules how to do some stuff so we can standardize them.
Like create a folder in /tmp to create files and so on.

Maybe can be helpful a shared file?

[Mte90]

As example:

• https://github.com/amber-lang/amber/blob/master/src/tests/stdlib/dir_exist.ab#L3
• https://github.com/amber-lang/amber/blob/master/src/tests/stdlib/change_owner.ab#L4

We need a function to create a custom temporary folder and one at the end of the test to remove it.

[Mte90]

I did a PR in the meantime #324

[CymDeveloppement]

Maybe the default output can be “Hello, Amber! » and if there is no output file, we could use this value?

[Mte90]

So the majority of tests should return "Hello, Amber!", if the .output.txt file is missing.
In this way we can simplify more.

It isn't a bad idea.

[Ph0enixKM]

I think that we can remove all the .output.txt by simply passing the result in the comment on top of the file. Here is an example:

// Hello There

let a = "Hello"
let b = "There"
echo a + " " + b

This can be easily parsed in the test module

[Ph0enixKM]

Wondering if now with the docs generator we can have a comment extractor, so get that information prefixed (so if there are comments in the tests that are just comments they don't create problems).

Comments // won't create problems even if they are documentation comments ///.

[Mte90]

We have the issue also that the output can be on multiple lines.

[Ph0enixKM]

Besides... that's just in the scope of the tests. Normal Amber code will work as usual. If you agree with me that we should do this @Mte90 then we can convert this discussion into an issue so that we can implement this.

[Ph0enixKM]

We have the issue also that the output can be on multiple lines.

I know. The algorithm works on multiple lines