Re: lynx-dev internal cache proposal

[Serge MUNHOVEN]

On Wed, Nov 18, 1998 at 04:43:04AM +0300, Leonid Pauzner wrote:
 >  
 >  >>       * compressed files?
 >  
 >  > Why not? :)
 >  cache uncompressed data or the original?
 >  
Or compress originals that don't already seem to be ? For a persistent cache
this could be a useful configuration option too.

Personnally I'm more interested in a short time cache. Kind of: I would like
to save a copy of the document I am currently looking at (or one of those
visited during this session) and don't want to transfer it once more. I was
already wondering if I could hack lynx to *replace* its current rendered
cache by a raw one, but I guess the proposal discussed in the recent days
is a far better idea on the long run.

As to non-HTML documents, they all seem to be already kept in temporary files,
though you do not have access to them ("e.g. interesting PostScript document
I'm looking at; would be nice if I could get into the downloader menu now, to
save it with a useful suggested name, since the document is already on my
disk").

I understand why results from POSTed queries should not be cached, but
shouldn't they still appear in the history and at least be accessible there
without reposting the request ? From what I understand, the trouble is the
arbitrariy length and contents of that kind of queries. If this could be
managed some how, those documents would be "historied" (trying to avoid
"cached") as well. I was not fully aware that there is "caching" for the
purpose of history and for the purpose of cache (or kind of mirroring). Thanks
to the previous contributors. :-)

But having all those documents in "history", shouldn't we leave the option to
get a prompt to : "revalidate" (and retrieve accordingly from the cache or the
net; this would be the default and the action taken straight away when this
prompt is disabled) or "redisplay" (I've been warned that the document may
be out-dated, but that's alright, just give me what you've already got).
There is already a similar choice (though between "resubmitting" and simply
"skipping" when moving back beyond the implemented history-"cache").

Just a few random thoughts (not even 5 cents).

Maybe the following triggers a few ideas or at least some interesting links :
  http://mnot.cbd.net.au/cache_docs/
though it already mixes up cache and history a bit (and the HTML does not look
that pretty with lynx, but that's an other story).

 - Serge