Re: last update

Abigail (abigail@tungsten.gn.iaf.nl)
Mon, 8 Jan 1996 22:59:41 +0100 (MET)


From: Abigail <abigail@tungsten.gn.iaf.nl>
Message-Id: <199601082159.WAA23789@tungsten.gn.iaf.nl>
Subject: Re: last update
To: www-html@w3.org
Date: Mon, 8 Jan 1996 22:59:41 +0100 (MET)
In-Reply-To: <Pine.ULT.3.91.960108231632.29035C-100000@hisar.cc.boun.edu.tr> from "!" at Jan 8, 96 11:36:19 pm

Imagine More!!! wrote:
++ 
++ Hi again!
++ 
++ Thanks for your all responses. 
++ 
++ > I seriously object against this approach, and object against
++ > Jon Wallis' suggestion of SSI's too. Both your and Jons suggestions
++ > use unnecessary bandwidth. Your image is generated by a CGI, and
++ > hence always "fresh", so it will never trigger a 304 not modified
++ > response (a lot of caches don't cache anything from cgi-bin anyway).
++ 
++ Mine was not a image but still a cgi-bin staff. I had wanted this "last
++ edited date" problem to be solved in an automatic way by the html itself,
++ without bothering the one who edits lots of html pages per day. (it is a
++ real problem if you have a forgetful mind like me :) )

Well, you have a way to put the html file on its final destination, don't
you? mv, cp, ftp, whatever. Simply write a wrapper for it, doing both
the move and the addition of the information. I think it's unprofessional
to let the reader (and the whole net) suffer for your laziness, specially
since you don't add any information. (Any server worth mentioning send
the modification time of the document anyway.)

++ > However, this is really not needed, as it is static information, which
++ > only needs to be updated when the document changes. Why not use the
++ > Perl script to put the last modified information *in* the html document?
++ 
++ Yes, a friend of mine just sent me an example of this kind of script. It
++ is not more than several lines of code. However it is away from being an
++ automatic solution :))
++ 								
++ Sorry, I don't have so much knowledge about working system of httpd, i.e
++ requests etc. Does my suggestion really cause so much bandwith problem? It
++ is not an image.. just 9-10 characters like : "Jan 07 1995" which may
++ appear in something like (somebody had suggested to use #exec): 
++ 			
++ <address>Last updated on <!--#exec cgi="/cgi-bin/update?foo.html"-->
++ </address>		
++ 					
++ Or, is it above all a matter of request which may have a standart form (?)
++ and does not depend on the amount of data -10 characters or a whole html
++ document- being requested?


It's not a 9-10 characters, it's the fact your whole document has to
be send again. Suppose your document is 50k, then the following is
not unlikely:

Reader visits your page, downloads 50k data, storing in her cache.
From the page, she goes to the next page, and the next.
After a few pages, she selects 'top' (pointing to your first page).
Her browser sends a 'get-if-modified' request. The server knows the
requested document has an SSI, so the requested document has to be
genereted *now* and is hence "modified", sending 50k again....
However, she has identical information in her cache.


Abigail