65636 - [RFE] unchecking "use pswd mgr to remember these values" should cause them to be forgotten

Status: VERIFIED FIXED

(SeaMonkey :: Passwords & Permissions, enhancement)

Description

[Max Christian]

Now that the "remember these values" checkbox is checked if a username/password
has been recalled (bug 40122), if the user unchecks the box and clicks OK, then
I suggest that the username/password combination that was recalled should be
forgotten (not recalled in future browser sessions).

To clarify, this is what I think should happen:
 1. Run Mozilla, login to a page with "remember" checked.
 2. Restart Mozilla.
 3. Go back to same page, login gets remembered.  Uncheck the box and click OK.
 4. View page OK.
 5. Restart Mozilla.
 6. Go back to same page, no login is remembered.

This is the way that Internet Explorer works, for what it's worth.  I think it
makes sense because (i) it's an intuitive way for the user to get a login
forgotten, and (ii) otherwise the checkbox has no function in the situation
where a login has been remembered and the user doesn't change the user/pass.

Comment 1

[Max Christian]

Attachment: proposed fix, using cvs diff -u

Comment 2

[Keyser Sose]

Changing to Enhancement and adding Patch keyword. Marking NEW, I agree with the
bug report btw.

Comment 3

[Stephen P. Morse]

I see the following problem with this patch:

You moved Wallet_GiveCaveat outside the "if checked" statement.  That means we 
might be giving the caveat even though we are not going to attempt to save the 
username/password.  That caveat should only be given when we know for sure that 
the password is actually being saved.

If Wallet_GiveCaveat is restored to its previous position, and assuming that you 
have tested this, then r=morse.

Comment 4

[Max Christian]

Attachment: updated patch with call to Wallet_GiveCaveat in the right place

Comment 5

[Stephen P. Morse]

r=morse for the corrected patch

Comment 6

[Brendan Eich [:brendan]]

I see no point in adding si_ForgetSignonDataFromBrowser when you could just call
si_RemoveUser.  The extra functional layer adds no value, as si_RemoveUser also
checks si_GetSignonRememberingPref() and returns false right away if that
predicate returns false.

/be

Comment 7

[Max Christian]

Attachment: version with no wrapper for si_RemoveUser

Comment 8

[Stephen P. Morse]

Great, that's much simpler.  r=morse.

Comment 9

[Brendan Eich [:brendan]]

Minimal, nice -- sr=[e-mail].

/be

Comment 10

[Max Christian]

Fix checked in 25/1/01, bug closed.

Comment 11

[Terri Preston]

Verified Win2k build 2001022007
Verified Linux build 2001022012
Verified Mac build 2001022013