Re: still a question

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Sun, 21 Apr 2002 23:59:42 +0200

On Sunday 21 April 2002 17:21, Songqing Chen wrote:

> yes, I found this. This way caused the new copy to be fully saved,
> even if the document is not changed. This seems wasteful.

Short of doing a full bitwise compare we have no way of knowing the
document is not changed.

> You mean the normal GET queries are few, so to save them again is
> not too much work?

Exacly.

> What for IMS queries? They are a lot. Is that because their replies
> are short or for some other reasons?

If IMS tells the object isn't modified we do nothing.

> ( And I looked into a trace file to find that what you said is true
> : the amount of IMS are much more than the normal GET, but I am not
> sure the reason: when I reload an URL, does the browser check its
> local cache to see if it has one, and if it has, it will send an
> IMS out? It seems somebody aslo mentioned the reload and "shift +
> reload" are different, but I am not clear about their result. )

When you do a normal "Reload" in your browser, the browser will send a
"GET IMS Pragma: no-cache" request, forcing the cache revalidation to
bypass the Squid cache.

Only Shift-reload (or control-reload) forces a "GET Pragma: no-cache"
request, forcing a new download of the object even if not modified.

Regards
Henrik
Received on Sun Apr 21 2002 - 16:03:08 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:15:17 MST