[Top][All Lists]

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

Re: lynx-dev Why doesn't lynx cache HTML source?

From: Leonid Pauzner
Subject: Re: lynx-dev Why doesn't lynx cache HTML source?
Date: Wed, 11 Nov 1998 13:10:05 +0300 (MSK)

>> > rules.  There have been bugs in the past relating to Lynx's current
>> > 1-document-in-1-rendering "cache", which is just about as simple as it
>> > can possibly get.
>> 1-document cache may be a first shoot:
>> probably kludge it to incoming socket stream
>> and check the cache somewhere before DNS search -
>> this is not the best way for speed (to use a complete internal circle
>> instead of Klaus's "internal link") but easy and clear.

> That might be OK.  I would prefer to see a full-blown implementation at
> the start, mainly because I think if we get the 1-document cache going,
> nobody will ever bother to do the full one...


Right, but I am in another position: my system have well turned Squid
so I will not get benefits from lynx cache unless "1-document" only.

It is not so hard to clarify the rules for "Visited Links" cache,
but takes a time (reading spec, discussion, new bugs, etc. etc.) -
this on the logical level, mostly implementation-independent.

And "1-document" case is highly lynx-specific. It can be expanded
easily if implemented onece.

>> > Even the simple 1-document cache has to interact with some caching

>> Really? Doesn't it enough to introduce one global flag in mainloop -
>> "reload internally" vs "reload as usual"?

> The problem is deciding what value to give that flag!  For instance,
> what should Lynx do if you read a document which expires in 5 hours,
> then walk away, come back 6 hours later, and ask for it to render

I mean the top history document for "1-document" case,
you will not probably look the same document for 6 hours,
but cache will be freed every time when you go to another document.
(It may only interfere with "Refresh= " directive,
but since it is not implemented properly yet - we should not worry about.)
But we should care for many-document cache, of cause.

> differently?  This is not the same as a document which was set to expire
> immediately -- Lynx *has* to assume that such a document is going to be
> stale some of the time, and let you do what you want.  But if it wasn't
> stale and now it is, maybe it should reload it.  Or maybe not.  That has
> to be determined.

reply via email to

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