🚀 go-pugleaf

> >
> >Aubrey: 
> >  for the benefit of some readers, or at least one (myself) 
> >could you please kindly post a reminder, what are Alfred and WTS?
> >
> >Could you kindly include just a few words of such reminder in your 
> >future postings (Alfred is...; WTS is...) ? Not everybody remembers 
> >your explanations from a few months ago, sorry.
> >
> >Wojtek
> 

Check out http://dejanews.dejanews.com/  
search for alfred & oberon

Alfred is a full text indexing program.  Aside from some file naming
problems, e.g. DB3Index.db3, &DB3Index.db3, and _DB3Index.db3
on Unixen, Intel/V4, and Intel/V3 machines, it works on all the environments
I have access to.  

The most recent version is on my home machine in a V3 environment,
and the power will be off at the office tomorrow and saturday.  I will
merge these all back together and have the web page updated soon,
perhaps by next Fri.

I am trying to get administrative approval to place a demo account
with a database on line so that interested, possibly non-oberon parties
can try it out.  The account already exists and is populated with data.

If anyone has an account that they are willing to publish guest login
particulars about, please give me a hollar.  I want to set up a demo
account, and then publish the access to oberon, <CHEMDISC@UMDD.UMD.EDU>,
and <HDESK-L@WVNVM.WVNET.EDU> readerships.

It may turn out to be faster and easier to implement the Windows/NT
demo.  That environment also exists and is online, and there is
a different administration involved, as well as a different operating
system paradigm.  This would support anyone with the Windows
Oberon environment and a WFW or NT environment.

One database, which cannot be made public, is the 18,000 email
messages sent to the various remark@{any}.cc.utexas.edu machines
since 8/93.  The total for these databases is about
130Mb
 35Mb
 3 Mb
for the three Relations involved in driving alfred.  

The indexing stats are interesting.  A 32 Mb SPARC 1+ running over NFS
takes about 80 seconds to index an "average" email.  A RS/600
takes about 40 seconds.  My 486-2/80, 12 Mb, 1+Gb, multiple
spindles NT machine took about 9.4 seconds, and the Pentium-100
32 Mb dell, 1 spindle scsi took 5.5 seconds to index an article.

For the 18,000 item database, ftp started at noon, Sun, Dec 3.
The job finished approximately 8 pm Wed.  This amounts to about
16 seconds / article, but includes system shutdowns, OS upgrades,
and other unrelated stuff going on.

I think this can be enhanced tremendously if I find an algorithm
to have a B-Tree that is balanced by "access depth" instead of
"absolute depth"  Many tokens, such as the funky things used in
the mail headers, are used 1 time.  Some other items, such as the
word I, and, the, are used extensively.  A simple change in the
file structure can make several hundred thousand file accesses
disappear.

I showed Alfred to the help-desk people.  They are parallel to me,
I manage the eHelp group.  I have noticed a curiosity that when
I show Oberon to Netscape aware people and mention "another
operating system" they have a meltdown and won't even look at it.

So, I didn't mention Oberon.  Or rather, I didn't say any of the
hot words like program, language, or operating system.  I just
told them I had a user-history database that they could drive
with mouseclicks.  They received it very enthusiastically that
first day.  I had the sense it would compete with their parallel
Netscape/sql environment.  I am going to continue to use the
Alfred application name with them, and observe.

A slightly older source version exists on 
http://ccwf.cc.utexas.edu/~mcintosh/Oberon/Alfred
This needs about 3-5 minutes work to make it be V3 compatible.
Apparantly, V3 uses the V2 convention to to a Viewer Textframe
setup, and V4 has diverged.  I think that I can converge these two
into one source.

One thing that I do wonder is whether there is a V3 --> V4 source
converstion tool.  WTS is the V4 --> V3 source conversion tool.

----
http://ccwf.cc.utexas.edu/~mcintosh
60 F high Dec 28.  Rain due real soon now.
-- 
http://ccwf.cc.utexas.edu/~mcintosh
12/28/95 --  60 F highs today.  Rain due yesterday, today.  
Not yet here.  Very dry, dewpoint and daily lows near freezing.
Broccoli, beets doing passibly well, leeks excellent!