auth: reworked pam auth, added pam_setcred

[xJayMorex]

Without calling pam_setcred, unlocking the machine doesn't also unlock GPG based secrets. Reworked CPam::auth to make it less repetitive.

[PointerDilemma]

In this new revision I don't like the refactor. Why not spend the extra lines to do the error handling?

About pam_setcred TBH I just can't find any proper documentation why this is required.
Hyprlock doesn't log you out or anything. Why do we need to extend the lifetime of some CREDs? I searched github and saw that a bunch of lockscreen related applications do indeed call it. But I still don't know why.

[xJayMorex]

In this new revision I don't like the refactor. Why not spend the extra lines to do the error handling?

You mean dropping the for loop and handling the steps individually, right?

About pam_setcred TBH I just can't find any proper documentation why this is required. Hyprlock doesn't log you out or anything. Why do we need to extend the lifetime of some CREDs? I searched github and saw that a bunch of lockscreen related applications do indeed call it. But I still don't know why.

If you aren't comfortable with secrets staying in memory while your screen is locked, you need to reset GPG before locking the screen (I have my keyboard shortcut set to gpg-connect-agent --no-autostart reloadagent /bye && hyprlock). You can unlock GPG in a separate step, it's just a matter of convenience. Display managers always unlock GPG during login, so I don't see why shouldn't all lockscreens do it also.

[spacefrogg]

pam_setcred is expressly there to also "maintain" (e.g. extend) credentials which is, surely, something to be done in a screen locker.

[PointerDilemma]

Ok I mean I guess doing pam_setcred is fine.

But why move the handle to the class. I think it was totally fine before no?