[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: address@hidden: lynx-dev caching after \ (was why reload etc?)]

From: Nelson Henry Eric
Subject: Re: address@hidden: lynx-dev caching after \ (was why reload etc?)]
Date: Sat, 11 Jul 1998 15:11:28 +0900 (JST)

> controlled by DEFAULT_CACHE_SIZE in lynx.cfg.  If your memory is so
> small, you could probably reduce that to (say) 5.  Or if the

Now at 4 on that particular machine.

I know I'm getting into another topic here, but they seem related
so I'll go ahead.  Actually just one file is enough to stop Lynx
dead in its tracks, and even worse, when leaving Lynx the memory
cleanup, i.e. time until the prompt comes back after q)uiting, can
easily take 5 minutes.  It varies, but any file over 300 KB will
certainly do it.

> It's a rendered version of the source.  The only way to get raw source
> is to download it.  Another way to put this point: Lynx isn't a file

As I thought.  Sorry, but I'm going to keep asking until I understand.
Does Lynx in its present state, then, ever have the _raw source_ in
hand to cache when fetching with `g' or hitting [CR]?  Seems to me
the answer is no.  Without the raw source in the first place, when
you toggle with `\' or change your character set or whatever, Lynx
has no choice but to go to the original source.  My present understanding
is that Lynx takes a stream of data and spits out a rendering.  If
the raw source is to be cached so that Lynx does not have to do refetching,
then between the receiving of the stream and the parsing, a middle man
will have to be put in to 1) write the stream somewhere (memory or
disk) and 2) then manage all of the cached URLs.  This "middle man"
is no trivial piece of code, but a veritable caching proxy server.

Then there are the people who simply want the "rendered source" like
what you get with a p)rint out after the `\' toggle.  This is not a
caching problem per se, but simply the way Lynx handles a certain
type of rendered document.  Seems to me it would be a pretty simple
hack.  Manually, I do it by d)ownloading the file and saving to disk,
then the `\' toggle is instantaneous because you're not going over
the 'net.  I could do this for the next ten years, and still spend
less time than it would take me to code the hack.
Unless I'm way off base here, it seems to me that caching the raw
source is not going to get done because no one with the skill to write
that code would bother reinventing the wheel.  Also, no one wants a
slow and bloated Lynx.  As to a hack to rename the rendered source so
that it could be kept in a cache of rendered documents along with the
parsed html of the same document, well it's been on the wish list for a
long time.

Now what I want is disk caching of the rendered documents so I can have
20 or 30 of them at hand, rather than 4 or 5.  Guess I'll have to wait
and hope that Fote will find fuN in Lynx again :-) 8-).


reply via email to

[Prev in Thread] Current Thread [Next in Thread]