User + key management

[eliocamp]

In thinking about the user verification it occur to me that currently there's no way of knowing is an user/password belongs to the webapi or the cds service. Should there be a service argument to wf_set_key() and the rest?

PS: You are free to tell me if I'm opening too many issues 😅

[khufkens]

Actually, this was implemented at some point.

Then I decided to remove it, for simplicity, as both the webapi and CDS use unique UID schemes (ECWMF uses email, CDS a random number). A week ago I tried to implement the CAMS atmospheric data and their api (http:https://macc-raq-op.meteo.fr/doc/Guide_Numerical_Data_CAMS_new.pdf).

Obviously, this one uses the email address again. 🤦‍♂️ So, yes... it would make sense to future proof things.

[khufkens]

Also, don't mind the issues, as long as you help out solving them 😄

[eliocamp]

Perfect!
Then in the service argument of the keyring I'll use something like ecmwfr_webapi/ecmwfr_cds.
This would be a breaking change, now that I think about it. What do you suggest?

[khufkens]

I think this should be documented well, then this shouldn't be too much of an issue.

[khufkens]

Actually, if you are still working on the readLines approach this could be a question that pops up. So people don't have to punch in the argument either. But I guess you had that covered.

[eliocamp]

Yes, that's when it came up.