expiry: when does expiry-wait start?

expiry: when does expiry-wait start?

[Kai Großjohann]

I think that there are a number of candidate dates when counting
expiry-wait might start:

* date header in message
* arrival date on local system
* last modification date of message
* when user hit `E'

If I'm not mistaken, Gnus uses different starting points, depending on
the backend.  I think it would be useful to think about what should be
the starting point to use, and how to implement it using the various
backends.  What do you think?

kai
-- 
~/.signature: No such file or directory

Re: expiry: when does expiry-wait start?

[Paul Jarc]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> I think that there are a number of candidate dates when counting
> expiry-wait might start:
> 
> * date header in message
> * arrival date on local system
> * last modification date of message

In the case of nnmaildir, these two are the same thing, and they are
the starting point.  This was just the easiest way.

> * when user hit `E'

The backend can't quite know this.  The backend can know when the user
entered and left the group, and notice that the user hit 'E' sometime
in between.  (If any backends try to do this, I imagine they probably
use the time of leaving the group.)

> If I'm not mistaken, Gnus uses different starting points, depending on
> the backend.

I don't know whether different starting points are in fact used, but
it is indeed up to the backend to decide how to do expiry.  Maybe all
backends just happen to all do it the same way.  (Or maybe not.)


paul

Re: expiry: when does expiry-wait start?

[Kai Großjohann]

On 27 Jun 2001, Paul Jarc wrote:

> I don't know whether different starting points are in fact used, but
> it is indeed up to the backend to decide how to do expiry.  Maybe
> all backends just happen to all do it the same way.  (Or maybe not.)

Okay.  I think all backends should do it the same way.  Somebody
(Lars?  ShengHuo?) should decide on what time to use, then it should
be written down somewhere, then all the backends need to be examined
and possibly changed so that they use the same time.

Or should it stay the way it is -- every backend does it its own way?

Opinions?

kai
-- 
~/.signature: No such file or directory

Re: expiry: when does expiry-wait start?

[Paul Jarc]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> Okay.  I think all backends should do it the same way.  Somebody
> (Lars?  ShengHuo?) should decide on what time to use, then it should
> be written down somewhere, then all the backends need to be examined
> and possibly changed so that they use the same time.
> 
> Or should it stay the way it is -- every backend does it its own way?

I think backends should do it the same way whenever it's not too hard
to do so - i.e., no gratuitous differences.  But each backend should
definitely document how it measures age, so users can take that into
account when deciding which backend to use, and so they won't be
surprised.  (I've just updated nnmaildir's documentation to include
this.)


paul

Re: expiry: when does expiry-wait start?

[Didier Verna]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) wrote:

> Okay.  I think all backends should do it the same way.  Somebody
> (Lars?  ShengHuo?) should decide on what time to use, then it should
> be written down somewhere, then all the backends need to be examined
> and possibly changed so that they use the same time.
>
> Or should it stay the way it is -- every backend does it its own way?
>
> Opinions?

        I have written a backend[1] which does a very peculiar kind of expiry,
so please let it be up to the backend. This doesn't prevent nnmail from
providing convenience functions that would do the usual stuff, though.


Footnotes:
[1]  I'll post the annoucement soon I think.

--
Didier Verna, didier@lrde.epita.fr, http://www.lrde.epita.fr/~didier

EPITA / LRDE, 14-16 rue Voltaire   Tel.+33 (1) 53 14 59 47
94276 Le Kremlin-Bicêtre, France   Fax.+33 (1) 44 08 01 99   didier@xemacs.org

Re: expiry: when does expiry-wait start?

[Kai Großjohann]

On 03 Jul 2001, Didier Verna wrote:

> I have written a backend[1] which does a very peculiar kind of
> expiry, so please let it be up to the backend. This doesn't prevent
> nnmail from providing convenience functions that would do the usual
> stuff, though.

Hm.  Of course, special cases might require special behavior.  I would
like to suggest the following:

All the standard email backends that come with Gnus should use the
same starting point for expiry.  The Gnus manual should document what
is the standard starting point for expiry, and say that backends
should clearly state if they are not following this standard.

kai
-- 
~/.signature: No such file or directory

Re: expiry: when does expiry-wait start?

[Simon Josefsson]

On Tue, 3 Jul 2001, Kai Großjohann wrote:

> All the standard email backends that come with Gnus should use the
> same starting point for expiry.  The Gnus manual should document what
> is the standard starting point for expiry, and say that backends
> should clearly state if they are not following this standard.

Yes.  So, to determine what the default should be, what date do various
backends use?  I'm guessing a little:

nnml: File system's modtime of article?
nnmh: Same as nnml?
nnfolder: X-Gnus header data?  Added when article saved in the nnfolder?
nnimap: Server-assigned date attached to each article.
	(Which usually equals the date when the article was copied
	into that mailbox.)
nndraft: never