Re: www @ nikhef

timbl (Tim Berners-Lee)
Date: Mon, 24 Feb 92 17:21:19 GMT+0100
From: timbl (Tim Berners-Lee)
Message-id: <9202241621.AA02717@ nxoc01.cern.ch >
To: W.vanLeeuwen@nikhef.nl (Willem van Leeuwen)
Subject: Re: www @ nikhef
Cc: www-talk@nxoc01.cern.ch
Willem,

Great to see the NIKHEF server on the web!  For others on the list,
see Willem's message below for examples of how to put together
www servers, including indexes, out of simple shell scripts.

I have put links to NIKHEF from our home page, and from the subject and  
organisation index.

There is one little bug in Xfind: the reference to //nic/ rather than //nic.nl./  
will work for those in the Netherlands but not for the rest of us.
(By the way, xfind is the name of a program by Bernd Pollermann which runs on VM  
... he might feel that the name is his.)

This is a neat server, particularly as it uses existing unix tools and data to  
provide a useful service.... keep up the good work!

	Tim BL

______________________________________________________________________
Date: Thu, 20 Feb 92 14:58:11 +0100
From: W.vanLeeuwen@nikhef.nl (Willem van Leeuwen)
Organisation: Nikhef-H (National Institute for Nuclear and High-Energy Physics)
Address: Kruislaan 409, P.O. Box 41882, 1009 DB Amsterdam, the Netherlands
Phone: +31 20 5920411, +31 2995 2499 (home)
Telex: 10262 hef nl
Telefax: +31 20 5925155
To: timbl@nxoc01.cern.ch (Tim Berners-Lee)
Subject: Re: www @ nikhef
Cc: a03@nikhef.nl

Hi,

There is a very preliminary version of xfind working on a very stupid set
of helpfiles.
These helpfiles are input to a VAX like help, so browsing with www
does not give very useful information, I only want to show that
the principle works.

If you want to try you may link NIKHEF into WWW with

	http://nic.nikhef.nl./user/a03/www/default/NikhefGuide.html

I now have 2 files which can be searched with keywords: the telephone
directory and this set of helpfiles.

The http daemon calls the script WWWsh:

WWWsh
=====

name=`echo $@ | awk -F? '{print $1}`
keys=`echo $@ | awk -F? '{print $2}`
name=`basename $name .html`
/user/a03/bin/$name.sh $keys

which may call Phone.sh or Xfind.sh

Phone.sh
========

name=$1
echo "<title> $name at NIKHEF</title>"
echo "<h1> $name</h1>"
grep -i $name /user/a03/www/default/phone.html

phone.html is a file which is generated every night from the finger
information on our central server.

Xfind.sh
========

name=$1
echo "<title> $name at XFIND </title>"
echo "<h1> $name</h1>"
#
# Do not forget to put 

#	pass	/user/a03/www/xfind/*
# in httpd.conf
#
cd /user/a03/www/xfind
files=`echo $name | /usr/lib/refer/hunt -Fn -Ty Index`
echo $files | sed -e "s/\///g" | awk '{for (i=1;i<=NF;i++) printf("<a  
href=http://nic/user/a03/www/xfind/%s>%s</a><p>",$i,$i)}'

The helpfiles are in a different directory, which has to be mentioned
in httpd.conf.

The index is made with the command

	/usr/lib/refer/mkey -w -f files | /usr/lib/refer/inv

files contains the names of the files to be indexed.
This is a rather crude approach, since the index contains a lot of keywords,
but again the first aim was to get something working.

I'll try to write down wy experiences with www in a more coherent way,
until then I hope this information is of some help.

Regards, Willem