Robert Collins wrote:
>
> 1) I can't fnd any cache flushing code for cached user credentials. Am I
> blind?
Probably not. It has never been considered an issue as the user base
usually is quite limited...
Size of the cache ~= n_users * (64 + 16 + 16) ~= 1MB for 10000 users
Users only get entered into the cache when their password is validated
as OK. Then they stay in the cache until some problem is detected..
(password cache validation fails because of invalid password or expired
cache entry).
> I want to do that in the NTLM tag because auth changes will be very hard to
> merge in from any other branch due to the existing extensive changes to
> modularise the basic/ntlm protocols. Any objections or strong reasons not to
> do it /not to do it in the NTLM tag?
The only reason might be to allow the two to go into HEAD at different
times, but I don't think that is a likely event.
So go ahead and "mess up" the whole authentication concept in the ntlm
branch even further. Having this separately is no big gain, and not work
the effort I think.
/Henrik
Received on Sun Oct 22 2000 - 06:54:49 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:52 MST