Subject: | |
From: | |
Reply To: | |
Date: | Thu, 28 Sep 2000 08:16:36 -0700 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Glenn/Cortlandt have made some good points. I'm excited
about the exchange, especially since they both have worked
for me at one time or another :-)
Please don't assume I am 100% forgiving of HP's corporate
e-speak that leaves the e3k out of the message. I was however
trying to bring the larger view into the mix. When you read
news stories about Sun, they are always linked with 'the servers
that serve up web pages on the web'. The impression is that
they are the only one doing this. HP appears to be making
a concerted effort to turn that perception around. Today's
e-press reviews of the recent Sun Ultra-Sparc III chip announcement
are critical of Sun's late delivery of the new chip into high-end
servers. HP and IBM have a golden opportunity here at the top end
and Carly appears to be capitalizing.
I wouldn't call the WSJ ad campaign worthless, but I would suggest
additional focus on the platform.
Here's a short list of investment opportunities which I think would
make a difference for the application development community.
1. A Java execution model with native threads, good performance
and integration with major subsystems (TurboImage, file-system)
and locking strategies so commercially viable applications
can be easily built without having to jump through hoops.
2. Training and development of materials to allow existing
and future e3k development teams to gain the core competencies
needed to use #1. [Note: see #10 below]
3. Build an application level transaction model that can be easily
called by the existing MPE NM languages. Make this part of FOS
and work extensively to develop materials, training, and migration
plans for customers wishing to convert from vplus. This transaction
model should be isolated from the source of the transaction such that
the application can easily mix custom built vb/delphi/java clients using
sockets with web transactions. This model should also include
authentication
and virtual session management, regardless of the source of the
transaction.
4. Adopt distributed object technology where appropriate to allow MPE
resident objects to communicate with NT/Unix/other objects over the
network. This would include support for DCOM. Over the years I've
suggested
CORBA support, but this is becoming less an issue.
5. Adopt SOAP and build a native interface for all MPE NM languages.
6. Develop an MPE COBOL with OOP support.
7. Deveop an MPE ODBC driver so that MPE code can be a client to ODBC
compliant data sources running on other systems, or on the same
MPE box. Make this part of FOS.
[Note: I think there may be solutions in this area from work done to
port Sybase Open Client, but I'm not sure]
8. Develop an XML parser/packager that can be accessed from MPE. Yes, there
are Java solutions, yes there are c++ solutions. But think a bit about
the average MPE developer and getting such support integrated into their
MPE applications and you start to see this need.
9. Can we get the e3k to e-speak? And fully integated with MPE applications
where appropriate.
10. Create a technical developer support team. This team would be charged
with
showing how, through examples and white papers, you can use the various
features/languages of MPE to develop world class applications. This team
would help define the 'best practices' in using the e3k for development
and would be a ready resource for 'how-to' in all aspects of application
development on the e3k. This team would be the developer advocate to HP
Development Labs for what is and not working on the e3k, based on input
from customer input.
[Note: we have some of this already, but its spread out over various
groups and not centralized.]
Duane Percox wk: 650.372.0200x608 fax: 650.372.3386
[log in to unmask]
www.qss.com
qwebs.qss.com
|
|
|