X-no-Archive:yes 

We are all too well aware that at one time, HP thought that the 3000 would
fade away. I wish I knew more of the details about what happened at an HP
World here in Boston, but I understand that some poor soul made the mistake
of saying something to that effect. The result served as a wake-up call to
HP that the 3000 was such a good platform that people who knew it and others
preferred it. Now they have committed to its renaissance. 

 
The 3000 not only has overcome Y2K, it has overcome sixteen bitness, and
will overcome thirty-two bitness, all the while remaining forward compatible
and backward compatible. HP seems incapable of marketing hype, at least WRT
the 3000. It is as they say, that any program written for the 3000 will run
today. And it will continue to run on sixty-four bit MPE. No one else can
come close to those claims. Even Windows has problems with sixteen bit code,
and NT can neatly stop you from running it, by refusing to run many sixteen
bit objects. 
 
Is the 3000 proprietary? And is it open? All the definitions I have read
seem to favor the agenda of the definer, somewhat reminiscent of
contemporary political rhetoric about what's good for America. In my work
with ftp, telnet, and rsh, it seems that each platform I have worked with
had it's own ways for implementation, or where less than friendly to systems
that could not do things their way, such that even 'open' systems are
proprietary. For those who care for the Gartner Group, they said that the
3000 is the most open system today. As for 'band-aiding' it, I am only aware
of a very small number of things that we wish the 3000 * could * do, that
are problematic, but I can think of none that are basically impractical or
simply impossible. The SIG chairmen could point you to 'wish lists' of
features that we want, and are very possible, but require time and money.
And you can take a look at the recent 'Oracle and other nonsense' thread for
an example of something that is generally regarded as a bad idea (adding a
'column' on this fly to an RDBMS 'table', actually a field to an IMAGE
dataset, while it is in use, by creating another file to contain that
column). Being able to do this in this way is a weakness, rather than a
strength. 
 
I was recently reading a chapter in Brooks's "The Mythical Man-Month", where
he deals with the problem of a system that starts out with a number of bugs,
then those bugs are found and fixed, and their overall number is at an all
time low. Then, various forces result in more subtle bugs being discovered,
and attempts to repair and enhance introduce new bugs. The system begins to
take two steps forward and another back, at best. I am not aware of the 3000
having reached this point; it seems to have successfully avoided this
problem in some measure for now. Furthermore, as one understands the 'tools'
of systems by layered models (client-server, OSI), or as isolated
components, or modules or objects, what is inside the 'black box' does not
affect other components, so long as the interfaces are accessible and work
as advertised. The analogy to token ring breaks down for this reason. 
 
I think little of the mindset that says, if it ain't broke, don't fix it,
having seem the definition of broke slide down the slippery slop into
unsalvageable. If it ain't broke, improve it. This history of the 3000
reflects this approach. It is a great tool, only in need of what it already
has, continuing improvement, but not repair. It ain't broke. 
 
opinions are mine, insert std disclaimers, your rights may vary from state
to state...
 
================================
All progress, all success, springs from thinking - [log in to unmask]
<mailto:[log in to unmask]>