No cache or not to cache

[Jo Gradeless]

There appears to be plenty of discussion in newsgroups on how to do caching
but nothing on why

I'm in some disagreement with one of my colleagues over the use of caching
of webpages.

He states that " I really do not expect to have to click my refresh button
on modern day sites. Years ago when amateur web building was all the rage it
was quite common to have to refresh."

and

" I do not think that our member societies should have the inconvenience of
remembering to refresh the site every time just in case they have missed
something. "

The pages change as required about once a month or less frequently. They are
not 'active' pages.

I dispute his arguments. I cannot see the point of reloading pages which are
unlikely to have changed in the course of the month. It seems to me that in
this case it is the responsibility of the person reading the pages to set
their own caching policy inside their browser.

I would welcome some discussion/reasons for and against caching.

Jo

[Alan J. Flavell]

Re: No cache or not to cache

On Mon, 30 May 2005, Jo Gradeless wrote:
[color=blue]
> There appears to be plenty of discussion in newsgroups on how to do
> caching but nothing on why[/color]

Not an HTML authoring issue. There are caches in browsers, and
cacheing servers out there, which will cache according to their
configurations. The only choice you have, as a provider of web
content, is to work with them nicely, or make a nuisance of yourself.
Understanding the issues is the first step to working nicely.

Tutorial at http://www.mnot.net/cache_docs/
[color=blue]
> He states that " I really do not expect to have to click my refresh
> button on modern day sites. Years ago when amateur web building was
> all the rage it was quite common to have to refresh."[/color]

Perhaps he too needs to understand the issues before reaching a
conclusion.

[Lāʻie Techie]

Re: No cache or not to cache

On Mon, 30 May 2005 14:37:50 +0100, Jo Gradeless wrote:
[color=blue]
> I would welcome some discussion/reasons for and against caching.[/color]

Caching came about because it is faster to read something off your hard
drive than to fetch it off the internet. Web servers should understand
the header request which asks for the meta-information about the requested
resource -- including its modification date/time. By only fetching
resources which have been modified since they were cached, we can save
bandwidth, time, and money.

The problem with caching is that there is no de facto way of specifying a
caching policy (ie how often do you expect the resource to change, or
perhaps it should not be cached at all). Once browsers and proxies have
unified, then caching will become more valuable.

HTH,
La'ie Techie

[Henri Sivonen]

Re: No cache or not to cache

In article <1117481945.35f d2979561461cd39 45fd75d2c126ef@ teranews>,
L®°°Æie Techie <laie@win_remov e_get_nospam_so lutions.com> wrote:
[color=blue]
> The problem with caching is that there is no de facto way of specifying a
> caching policy (ie how often do you expect the resource to change, or
> perhaps it should not be cached at all).[/color]

There is, however, a de jure way. See RFC 2616.

--
Henri Sivonen
hsivonen@iki.fi

http://hsivonen.iki.fi/

Mozilla Web Author FAQ: http://mozilla.org/docs/web-developer/faq.html

[Alan J. Flavell]

Re: No cache or not to cache

On Mon, 30 May 2005, L??ie Techie wrote:
[color=blue]
> Web servers should understand the header request which asks for the
> meta-information about the requested resource --[/color]

A "HEAD" request, you mean? I'm sure they do (support for it is a
mandatory requirement of the web server specification), but that isn't
how cache-friendly retrieval is done.
[color=blue]
> The problem with caching is that there is no de facto way of
> specifying a caching policy[/color]

Perhaps you could be persuaded to read the widely-cited tutorial on
this topic, and then come back with some more-specific point.

http://www.mnot.net/cache_docs/

What you just claimed is simply untrue: the facilities are not only in
the RFCs but are, in everyday practice, implemented in any decent web
server (such as Apache), and have been for years already.
[color=blue]
> (ie how often do you expect the resource to change, or
> perhaps it should not be cached at all).[/color]

In practice, it works. The key missing part seems to be learning how
to do it. Which is doubtless why Mark Nottingham wrote (and continues
to update) that fine tutorial. I'd recommend to make use of it.