Made format constants public

[lewisfish]

Hey,
This resolves #570.
Pull request makes the format constants from stdlib_io public.

Let me know if I need to do anything else as this is my first pull request, so not 100% sure of the protocol.

Lewis

[awvwgk]

Thanks for starting this. Have a look at https://github.com/fortran-lang/stdlib/blob/HEAD/doc/specs/stdlib_io.md to add a description of the constants which should become part of the public API.

[awvwgk]

Another point for discussion is how the format specifier should be exported. Do we want to be able to use it in a context like this:

use stdlib_io, only : FMT_REAL_SP, FMT_COMPLEX_SP
implicit none

print '('//FMT_REAL_SP//',1x,'//FMT_COMPLEX_SP//')', 1.0, (2.0, 3.0)
end

I wonder whether the *(...) repeat will mess this up.

Further, what about non-advancing output, some format specifies have a trailing skip (1x for int and real), other don't (complex). Should we include an explicit break for such case, i.e. (es24.16e3,:,1x)?

[lewisfish]

Thanks for starting this. Have a look at https://github.com/fortran-lang/stdlib/blob/HEAD/doc/specs/stdlib_io.md to add a description of the constants which should become part of the public API.

Added now, let me know if you have any comments on the format of this as wasn't too sure what to copy format-wise from the other examples in the docs.

I wonder whether the *(...) repeat will mess this up.

Further, what about non-advancing output, some format specifies have a trailing skip (1x for int and real), other don't (complex). Should we include an explicit break for such case, i.e. (es24.16e3,:,1x)?

Is there any reason the complex doesn't have the trailing skip? It would make more sense if all the formats had the same output

[milancurcic]

Good points, I think we should export constants without the *(...) and the 1xs, and let the user adapt them for arrays if needed.

Is there any reason the complex doesn't have the trailing skip? It would make more sense if all the formats had the same output

I think I implemented this and I don't remember why it is like this. It's possible that I did it by mistake and the tests were insufficient to make it an issue.

[lewisfish]

Good points, I think we should export constants without the *(...) and the 1xs, and let the user adapt them for arrays if needed.

Is there any reason the complex doesn't have the trailing skip? It would make more sense if all the formats had the same output

I think I implemented this and I don't remember why it is like this. It's possible that I did it by mistake and the tests were insufficient to make it an issue.

Great, I'm happy to implement this.
save_txt uses these constants, the easiest way I see to implement this would be to remove the * and 1x from the format constants then concat them in the required places in save_txt. Does that make sense?

[milancurcic]

save_txt uses these constants, the easiest way I see to implement this would be to remove the * and 1x from the format constants then concat them in the required places in save_txt. Does that make sense?

That seems fine to me. 👍

[lewisfish]

Okay, updated the code and docs. Let me know if anything needs changed or adjusted.

[milancurcic]

Thank you!

[awvwgk]

Thanks for sharing. Looks good to me.

Could you add a note about this change in https://github.com/fortran-lang/stdlib/blob/HEAD/CHANGELOG.md as well?

[lewisfish]

Thanks for sharing. Looks good to me.

Could you add a note about this change in https://github.com/fortran-lang/stdlib/blob/HEAD/CHANGELOG.md as well?

Done. Thanks for all your and @milancurcic help in getting this done!

[jvdp1]

Thank you.
Small questions: is there no tests?
Also, I think that these will not be documented in Ford docs. Could you add some comments for each format (or at least one general comment) with a link to the specs, please?

[lewisfish]

Thank you. Small questions: is there no tests? Also, I think that these will not be documented in Ford docs. Could you add some comments for each format (or at least one general comment) with a link to the specs, please?

With regards to tests. There are no explicit tests. However, there are some implicit ones within the save_txt tests since save_txt uses the format specifiers. Is that sufficient?

Added some comments for each specifier. Let me know if I've done the documentation wrong as I've not used FORD before.

[jvdp1]

Thank you @lewisfish . It looks good to me!

[milancurcic]

Thank you @lewisfish and reviewers!