should auth-source cache false results?

should auth-source cache false results?

[Ted Zlatanov]

It's pretty common that a search will return no results and :create is
nil, so we return nil from `auth-source-search'.  Do we want to cache
those cases?  Right now they are not cached.  It could be an option,
too, but you can argue both ways for the default.

Ted

Re: should auth-source cache false results?

[Ted Zlatanov]

On Thu, 17 Feb 2011 06:06:16 -0600 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> It's pretty common that a search will return no results and :create is
TZ> nil, so we return nil from `auth-source-search'.  Do we want to cache
TZ> those cases?  Right now they are not cached.  It could be an option,
TZ> too, but you can argue both ways for the default.

Thinking about this, we should at least cache the absence of results if
the source file for the netrc backend has the same time as before.

Ted

Re: should auth-source cache false results?

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> Thinking about this, we should at least cache the absence of results if
> the source file for the netrc backend has the same time as before.

Sounds good to me.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: should auth-source cache false results?

[Ted Zlatanov]

On Thu, 17 Feb 2011 14:41:38 -0800 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> Thinking about this, we should at least cache the absence of results if
>> the source file for the netrc backend has the same time as before.

LI> Sounds good to me.

In practice this is, I think, a bad idea (I tried it out).  If the user
adds entries externally, he will be mystified and won't necessarily know
about auth-source-forget*.  So instead I cache all netrc searches now.

If you think there's a better way to do this, let me know.

Ted

Re: should auth-source cache false results?

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> In practice this is, I think, a bad idea (I tried it out).  If the user
> adds entries externally, he will be mystified and won't necessarily know
> about auth-source-forget*.  So instead I cache all netrc searches now.

If the .authinfo file changes, then auth-source should forget those
entries automatically, shouldn't it?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: should auth-source cache false results?

[Ted Zlatanov]

On Wed, 23 Feb 2011 00:33:31 -0800 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> In practice this is, I think, a bad idea (I tried it out).  If the user
>> adds entries externally, he will be mystified and won't necessarily know
>> about auth-source-forget*.  So instead I cache all netrc searches now.

LI> If the .authinfo file changes, then auth-source should forget those
LI> entries automatically, shouldn't it?

Yeah.  I must have screwed it up.  But now the other problem is that
false results and no results look the same, being nil.  How do I store
that semantic distinction?

Ted

Re: should auth-source cache false results?

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> Yeah.  I must have screwed it up.  But now the other problem is that
> false results and no results look the same, being nil.  How do I store
> that semantic distinction?

Hm...  is there a difference?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: should auth-source cache false results?

[Ted Zlatanov]

On Thu, 24 Feb 2011 19:05:32 -0800 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> Yeah.  I must have screwed it up.  But now the other problem is that
>> false results and no results look the same, being nil.  How do I store
>> that semantic distinction?

LI> Hm...  is there a difference?

With the brand spanking new `password-in-cache-p' I can distinguish
between a present key with a nil value and an absent key.  It's pushed
together with the auth-source.el changes to use it; please let me know
if you have problems.

Ted