Bizarre Cache Implementation

Bizarre Cache Implementation

[Arnd Kohrs]

Hi All,

is there any particular reason for the current awkward cache
implementation? 

By awkward I mean that it should be implemented like gnus-agent, but it
isn't.  In fact, it seems to me that it just a lower quality implementation
of nnml.  For example gnus-cache does not deal correctly with
compressed files (jka-compr), while nnagent (based on nnml) does.

It seems to me that gnus-cache.el should be rewritten thus it merely
interfaces with one (or multiple) storing backend like nnml or nnfolder.  

The advantages would be fewer unwanted features (like the one above) and
more flexibility as to where one actually stores one's cached files
(even shared caches could be possible if the backend was nnimap (IMO a
very useful feature for slow web-based discussion groups)).

I might oversee important issues here.  Also there is probably a good
reason why the current implementation is like it is.  Maybe somebody is
already working on this issue?

Arnd.
-- 
The Active WebMuseum: Your personalized access to art paintings.
 Visit now ->  http://www.eurecom.fr/~kohrs/museum.html

Re: Bizarre Cache Implementation

[Stainless Steel Rat]

* Arnd Kohrs <kohrs@acm.org>  on Fri, 23 Jun 2000
| By awkward I mean that it should be implemented like gnus-agent, but it
| isn't.

Consider that gnus-cache predates gnus-agent by at least a year.
-- 
Rat <ratinox@peorth.gweep.net>    \ Ingredients of Happy Fun Ball include an
Minion of Nathan - Nathan says Hi! \ unknown glowing substance which fell to
PGP Key: at a key server near you!  \ Earth, presumably from outer space.

Re: Bizarre Cache Implementation

[Lars Magne Ingebrigtsen]

Arnd Kohrs <kohrs@acm.org> writes:

> The advantages would be fewer unwanted features (like the one above) and
> more flexibility as to where one actually stores one's cached files
> (even shared caches could be possible if the backend was nnimap (IMO a
> very useful feature for slow web-based discussion groups)).
> 
> I might oversee important issues here.  Also there is probably a good
> reason why the current implementation is like it is. 

gnus-cache predates nnoo, which is why it is like it is.

Storing the cache in a backend of your choice sounds OK to me, but the
current cache implementation is nice and fast, so I see no pressing
need to rewrite it...

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