On Jan 16, 2006, at 9:40 AM, Chris Richardson wrote:
It was while working for Shared Medical Systems that I learned the
MUMPS
language (1978). It amazed me that the MUMPS side of the house had
a mean
time to repair of about 15 minutes for a remote site. Their Cobol
side of
the house was lucky to get 72 hour turn around. Changes on the
MUMPS side
of the house were quick and painless (no evident recompile and
link). The
economies of scale are very evident.
You mean back when you could only compile by submitting batch jobs
overnight? If that is the only argument for using MUMPS, we're in
trouble, because languages that support (if not require)
interpretation abound, and in almost all cases, recompiling is
relatively cheap. Of course, none of this means that MUMPS wasn't
ahead of its time or that the advantages you describe aren't real.
The Department of Defense has their Composite Health Care System
(CHCS I)
which was developed from the VA's DHCP (the early name for VistA.
In the
compute-off that went into the DoD's effort to use the VA code
base, there
were 4 vendors involved. Only one vendor needed to use the VA code
base
(SAIC). Two of the 4 vendors dropped out after spending the $25
Million
allocated for building an entry into the compute-off. It was down
to two
vendors, SAIC and MacDonnel-Douglas doing a more conventional
approach. The
MacDac entry was 67% percent functionality and their bid was $2.6
Billion.
The SAIC entry into the competition was 98% functionality and the
bid was
$1.01 Billion. In less than the10 years of the runofthecontrat, the
project
was fully operational and brought in on time and on budget.
That is certainly impressive, and even newer technologies have tended
towards large, complex infrastructure with all the resulting
problems. I think there's a growing backlash, though.
MUMPS is an easy language to learn,
Is it? This is where I don't think I agree. The language is not
terribly complex (neither is C), but does that make it easy to learn?
Perhaps it is easy to learn to write programs in MUMPS (not quite the
same thing), but the same is true of Basic, Python or Scheme.
and it has been easier to train
knowledge-area specialists to program in MUMPS than to try to
capture the
requirements and have trained programmers do the work in abstraction.
I think the idea of teaching knowledge-area specialists to program is
interesting, and something that needs to be explored more fully than
it has been. But there's a lot more to being a good software
developer than knowing a language (or two) and being able to put
together simple programs.
The
code becomes the medium for the specification. VistA is a model
which
helps to standardize these requirements (at least a bit) while
providing a
consistant interface to the user and the application.
I would turn that around: instead of using general purpose
programming languages as a specification tool, we need to learn to
develop specifications that are directly implementable.
===
Gregory Woodhouse
[EMAIL PROTECTED]
"The most incomprehensible thing about
the world is that it is at all comprehensible."
--Albert Einstein (1879-1955)
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
_______________________________________________
Hardhats-members mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/hardhats-members
