Closed MichaelJW closed 3 years ago
Thanks! Looks great! One quick question before the details. Does Windows actually store the username? The docs say
If the credential Type is CRED_TYPE_GENERIC, this member can be non-NULL, but the credential manager ignores the member.
I hadn't noticed that line in the docs. You're right, it suggests that Windows shouldn't be storing the username. However, my tests (on Windows 10, FWIW) show that it does:
Strange...
OK, I checked on Vista, which is the oldest versions that current R supports, and that has a username field as well, so we should keep it.
Hey @gaborcsardi, anything required from me here? I have more time on my plate from this week on now that lockdown's eased a bit and schools are back on :smile:
Thanks much, it looks awesome! I decided to create a lean lower level keyring package, that implements the non-portable APIs as well, and I'll use this code there: https://github.com/gaborcsardi/oskeyring (Only the macOS code is new there, Windows is next.)
Eventually this package will just call that.
Sounds great, thanks - I'll keep an eye on that package!
Thanks again! I will not merge this here, because the oskeyring package has this functionality now. Hopefully you can use that for your use case. Please open an issue otherwise.
Relates to this issue: https://github.com/r-lib/keyring/issues/84