Read author information from a file in the repository

[armish]

This is in conjunction with issue #4 and also related to issue #6.

It is a really good idea to fetch the names automatically from GitHub (see issue #4), but I think we can try to come up with a best-practice where every project has an AUTHOR file in the root folder and the authors of the projects can be responsible for maintaining the file.

People are already using legacy files such as README and LICENSE; I don't think it will be problem for many of us to include an AUTHOR file in the repository as well.

This will also take the burden of hitting GitHub server and resolving username/names off of the submission system and it will further allow better customizable list of names and orders in the long run.

[robldavidson]

Is it better to have multiple files within the repository (arguably easier for human understanding) or to have one catch-all file with all metadata in an arguably more machine readable format e.g. XML, JSON?

If the authors had been a bit slap-dash and not created all appropriate files/fields in their repository, would the DOI submission system be able to collect the missing metadata (easy) and then add the files/fields to the GitHub repository (???)

[RichardLitt]

I agree an AUTHOR file is a good idea. Is there a way to automatically generate it from a list of contributors to a repository?

I think having multiple files is better. README files have a specific function on GitHub, whereas AUTHOR doesn't. I don't think it's too much to ask the users to include each of the three files needed. Automatically collecting missing information or throwing a flag if it doesn't exist should be part of the DOI submission system.

[ThomasA]

Which format should such an AUTHOR file have and what information should it contain? Names, obviously, but other things could be ORCID (if available), and possibly author roles?

[ggorman]

+1 ORCID support. This is pretty much the only game in town for standardising identity. Emails and even names change (e.g. marriage).

[robldavidson]

Agree regarding ORCID. Uptake is growing among researchers but it may not be so popular across the board - we need to see.

What is a minimal information standard for author? Name is too ambiguous, emails change too regularly, ORCID is 'just another social network' (although I love it and hope publishers/funding bodies demand its use).

Perhaps 'name'+'one social network account'? For some people, their Facebook, Github or LinkedIn will be as stable/unambiguous as ORCID.