SLAC WWW-Tech Mtg 10/897

Attendees: Tony Johnson, Les Cottrell,  John Halperin, Laurie Gennari, Joan Winters, Charlotte Hee, Ruth McDunn, Dennis Wisinski, Bebo White, George Crane, Glen Biggus, Andrea Chan, Bob Cowles, Paul Raines, Mike Wendling, Richard Dominiak..

Agenda:

Meeting Maker (MM)

The request to buy on last year's budget for ~$45K was rejected last year.  Chuck Boeheim
got another bid from MM for ~$16K for 500 licenses, the bid ran out on September 31st.
A justification for purchasing MM has been sent to Steve Williams, he has been busy with budgets etc. It does now support Solaris. There is no money in the SCS budget for FY98 to cover MM. Tony Johnson has a calendaring server from Microsoft (based on Exchange) that he is supporting for SLD. Contact Tony if you are interested in using. The AltaVista Forum appears to be more of a personal calendar system than MM.

Using Java at SLAC - George Crane

George introduced Glen Biggus who is developing Web applications for PoepleSoft.  They would like to deploy more Java based applications, and they wondered what, if any, the policy at SLAC is towards Java. Earlier on there were concerns about Java security, and though the Netscape, Sun Java implementations appear OK, John Halperin is concerned that MSIE implementation does not have a good reputation for its security yet. At the moment it is an individual decision as to whether to allow Java applets to run from their browser. Though it is fairly easy to configure the browser to accept or reject Java applets, it would not be realsitic to change the setting many times during a session as one looks at different pages. Tony warned that applets may not work across all browsers (e,g, stuff developed in JDK 1.1) will only work in the most current browsers, so careful testing may be needed. The main target for the BSD Java applets would be the accounting folks who mainly have WNT.  The consensus was for BSD to go ahead and develop the Java applets. People with Macs who may not have enough memory to run the most current browsers (which support the latest Java) may be out of luck.

Schedule for new WWW Unix Server - Dennis Wisinski

They hope to bring up the server for pilot use in the next 2 onths, and then to go full production early next year. They will run the Enterprise server version 3. Charlotte raised the question of why we chose the Ntscape server (as opposed to say Apache).  The decision was made over a year ago, and things may have changed from that time. A lot of work went into making the CERN server realiable (killing runaway scripts etc.) which may require work to replicate under Netscape or may be embedded in the standard server. There are also concerns how some things (e.g. slaconly) may be supported (if at all) under Netscape.  It may be necessary to continue some of the services with the old CERN server.

Running cron jobs on WWW servers - Paul Raines

For supporting Hypernews Paul runs a small (<15mins) job to index Hypernews. He runs it on WWW so he does not need to worry about the token issue. It was agreed that it is a reasonable request and asked Paul to talk to Bebo.

Encryption Keys

Damien Pierce requested support for 128bit public keys.  This is very secure, but sounds more security than we really need. Tony believes that it automatically drops to 40bits if the transaction is international. Since SLAC has a lot of foreign collaborators who can only use 40 bit encryption if most of SLAC is using something else we may not be testing things correctly. It was agreed that in lieu of having a strong enunciated need for 128bit key support we should leave things as is.

Statistics & robots.txt

Many of the hits in the statistics log come form Web search engines.  Should we increase the robots.txt to indicate areas that the search engines should avoid.  The SLAC Harvest engine avoids a lot of SLAC Web space. Mike Wendling will look at extending the Harvest configuration file for robots.txt

Status of WWW6CONF

No progress, nobody volunteered to make migration.  Christine Quinn of camous is talking to Tim Torgenrud to see if he could do it.

IP Number instead of Domain Name - Joan Winters

PCD has a hard coded IP address for its server. It is on port 8080 under pcd-one.slac.stanford.edu:8080. They claim that they want to use an IP address (134.79.113.153) since it avoinds the name server and so is more reliable. The SLAC name servers are designed to be highly reliable (on UPS, automatic paging in case of problems etc.) so very little additional unreliability added through the name server.  The name server provides extra flexibility, and also is easier to use/remember  than an IP address. It was recommended that the PCD server is referenced by it name.

Aliases for www2 - Joan Winters

The BIS group would like to have an alias for the www2 Web server called www-bis. The intent is to allow migration of the BIS web pages to a separate server at a later date.  This alias was recently set up.  They would like to start their URL tree one level down in the file structure so as not to have to repeat the bis, e.g., http:www-bis/something.html rather than http:www-bis/bis/something.html (current state).

In addition, it might be useful to create two other www2 aliases, www-user and www-group, for the user space analogous to public_html  and /afs/slac/www/grp space on www1.  We plan to move the users off to their own Windows NT server as soon as feasible.  Providing for the separation of groups from topical Web space may also prove useful in the long run.   There was a short discussion of alternate names www-users and www-grp.  Since www2 is not UNIX, no one preferred www-grp over www-group.  Some preferred and no one strongly objected to www-user over www-users.
It is consistent in number with www-group.

Tony pointed out that the www1 aliases did not prove useful because users incorporated them in their own URL not according to the design, e.g., they used www-spires to reference the SLAC Home Page.  However, he thought it may be possible through vitual hosts to enter a Web space below the root  with only the one IP number.  This could provide different URL per alias as George wanted and would generally make aliaises more useful. ( According to Richard Dominiak, the Netscape server provides these features.)  Tony, Dennis, and Joan will investigate whether we can do this on the www2 IIS (Microsoft) server.

Les Cottrell