Re: Refactor elpa/oauth2.el to use plist over cl-defstruct

[Adam Porter]

Hello,

> I have some more free time so I was looking into my oauth2.el again. I 
> want to ask if it would be desirable to refactor the library to use 
> plists in place of `cl-defstruct`? This would make more sense as the 
> storage backend used is plstore, which uses plists as its native format.

Instead of refactoring the library to use plists, have you considered 
using a different storage backend, such as persist.el or 
multisession.el?  It needn't be limited by plstore's design, and 
swapping the storage backend would probably require fewer changes.

> I have also read that the use of cl-lib is discouraged for various reasons.

Be careful with the use of the passive voice.  It's very easy for 
misleading rumors to be spread that way.  ;)  The questions are of who 
discourages such use and why.  And as others have said, it isn't the case.

> My overall goal is to make oauth2.el integrate with auth-sources, and 
> properly hook into functions like `auth-info-password` to refresh (if 
> needed) and supply the token, in hopes to better integrate with nnimap 
> in gnus.el.

Do you mean `auth-sources` or `auth-source`?  I can only find `auth-source`.

And with regard to using `auth-source` for storing and retrieving 
tokens, I'd point out the problems I encountered when trying to use it 
that way in the past:

https://old.reddit.com/r/emacs/comments/8lvda6/is_authsource_from_the_dark_side/

AFAIK those issues have not been resolved, and until they are, ISTM that 
it's not suited to the task.  As I've mentioned elsewhere, Emacs could 
benefit from a "new" (i.e. not that it necessarily needs to be 
implemented from scratch) secrets library.

My two cents.

--Adam