Python and Database

[Noelle QUEMENER]

Hi,

I am working on a new project which aim is
- to centralise data in an Oracle Database,
- to do some calculations (based on models),
- to visualize these data on diverse client PC (heavy clients).
The OS environnement is Unix.

The context of the system is
- in case of crash the user mustn't lose his context and can work in a
degraded mode
- high transparency to failures
- high availability of the system (10 minutes allowed between 7h and 14h -
1 minute allowed between 14h and 15h30
and less than 1minute between 15h30 and 16h30)

The techical architecture is not yet chosen but experts advise us to use a
Cluster Weblogic application server combined to a Cluster
version of Oracle and java developement on the clients (heavy clients)
What I know is that java is not a performant tehnology in terms of speed of
execution and Cluster softwares are extremely expensive.
I discovered Python not so long ago and to what I saw it is surprising how
compact/performant it is and how easy it is to use. I would like to
propose Python by I have no argument to support my suggestion. Can someone
help me?
My further questions are:
- we don't want users to connect directy to the database but we want them
to use an interface. Can such an interface be deployed in Python,
knowing our time constraints?
- as we want a high availability of the whole system what kind of solution
could we use?

Thanks,
Noelle Quemener

 

[Patrick Useldinger]

On Tue, 14 Oct 2003 10:59:58 +0200, "Noelle QUEMENER"

You have a lot of constraints, and I suppose that you didn't mention
that you also have very short deadlines ;-)

There are 2 issues here:

- 1: Architecture -
this is critical for your project, and more or less independant of the
programming language you want to use. Decide this first. If you want
this high level of availibility, clusters are probably a good way to go.

- 2 : Language -
Basically, I am tempted to say that you should use whatever your
programmers know best. The application you describe can probably be
written in any language. Java is a sure guess though, because it has a
lot of support.
For speed of calculation, consider writing the number crunching code in
a compiled language.

Using a new language in a new project is IMHO only suitable if this is a
pilot project whose aim it is to evaluate the language. Don't make the
mistake many large corporations do, i.e. expect a new technology to be
the silver bullet for a difficult project. I have seen some of these,
and they all produce bad results, if any at all.

What is your company's politics? In our company, you don't just "try
another language", because this raises a lot of issues about programmer
competence, maintenance, ...

This is not a testimony againt Python. Python is my preferred language,
but I believe every language needs to be "played around with" first
before using it in a business project.

-PU