Making a simple file server for unix

timbl@www2.cern.ch
Tue, 26 May 92 15:55:11 GMT+0100


Date: Tue, 26 May 92 15:55:11 GMT+0100
From: timbl@www2.cern.ch
Message-Id: <9205261455.AA00465@ www2.cern.ch >
To: John O'Neall <JON@frcpn11.in2p3.fr>
Subject: Making a simple file server for unix
Cc: www-talk@nxoc01.cern.ch

> Date:         Tue, 26 May 92 09:16:33 EST
> From: John O'Neall <JON@frcpn11.in2p3.fr>

John,

Thanks for your mails asking about how to make a simple hypertext  
list of files. 


Let me take your model of the three layers. I sugest that the 1st  
layer

>helppage (1st-layer):  various info and pointers to local
>         info that won't change much as well as to WWW proper.
>         Among other things, it'll point to a [...]

you write by hand.  You can read the HTML documentation in    
http://info.cern.ch/hypertext/WWW/MarkUp/MarkUp.html
or you can just copy another node which looks like what you want.   
(Get its address and then read it with www -source address). Or both.

You might want to put pointers to the general subject index, and the  
index of HEP sites.  Let me know when you set up a server so that I  
can put in on the list!

For the second layer,

> 2nd-layer list:  this is the thing that will point to ALL the
>    local flat files.  Then I need an automatic (cron'd) procedure
>    to regenerate the list every time I add more flat data
>     files (or periodically).


This is easy: you generate it with a little shell script. How about  
the one appended to this message?  It generates a hypertext menu of  
files in a set of directories (passed as arguments). You can massage  
it to put in your own title, heading, etc. If you can generate some  
human-readable description of each file, so much the better.  You  
could run this as a cron job into an Overveiw.html file which you  
then publish with httpd, or run it from a server daemon script so  
that the list is generated fresh for every read, and always  
up-to-date.

The third layer is your set of files.  If you publish these with a  
server script, you can add to each one a title (if you can generate  
it) and maybe a link back to the list of other documents befre you  
just send the plain text file.

You could read some of the tips on the web about ettiquette, etc.

>By the way, if such a list already exists in W3 somewhere, I'll be
>happy to point to it too.  And then there's no reason why our server
>shouldn't be accessible to the rest of W3.

It would be a great addition to the web!  Let me know when it's up  
and I'll put pointers to it from some overview documents. (evn at the  
experimental stage, I'll mark the links "experimental" if you like).

>Tim, any chance of getting the hypertext editor for something other
>than Next one day?  I realize that was the simplest thing to develop
>it on, but I suspect it's not the commonest workstation in HEP.

It's on the "hit-list" and we're lookig for volunteers. Maybe the Mac  
browser will be the first to be also an editor. Or maybe one of hte X  
browsers.  I agree the NeXT is not the most common platform! (Though  
it is neat ;-)

>Excuse me for rushing, but I'm supposed to present this project to  
the
>HEPIX-F meeting in Paris on Tuesday and this weekend will be a long  
one
> (at least, in France).

Good luck! Mail cailliau@(I'm off to the US -- so I'm rushing too!)

Mail me or www-talk@info.cern.ch if you have any problems... 


If you have smart ideas, circulate them to this list too.

	Tim

__________________

#! /bin/sh
# Generate hypertext menu from directory list
echo "<TITLE>Information under: $*</TITLE>"
echo "<H1>$*</H1>"
# If there is a README file include that as plain text
if [ -f $1/README ]; then
    echo "<XMP>"
    cat $1/README
    echo "</XMP>"
fi
# Now generate a list of links to files
echo "<DIR>"
for dir in $*
do (
    cd $dir
    for file in *.html *.txt
    do 

	echo "<LI><A HREF=./$dir/$file>Title of $file</A>" 

    done
    )
done
echo "</DIR>"