Need advice on choosing the technology for intranet based application

[Ian Bicking]

On Wed, 2002-09-18 at 17:30, Nenad Propadovic wrote:
> Hello pythonistas,
> I'm about to write an application related to injury prevention in big
> industrial companies. It's supposed to be intranet-based (because the admins
> are reluctant to install it on x (where x>>1) clients). So: the client is a
> web-browser. I still don't know whether I will get a server on my own, or
> I'll have to fit into an unknown web server environment, both options being
> realistic.
> By now it seems the logic of the application will be simple. It's going to
> be a relational database with several tables, a modest number of users with
> write priviledges, some more with read priviledges only. Performance is not
> a big issue.
> The GUI has a part I consider complex: I have to present some statistics in
> form of charts or graphs.

I would encourage you to consider the graph-creation process to be
separate from the web interface.  You can program and debug this outside
of a web environment (e.g., just creating graphics files in a
directory), and then plug it into whatever environment you want -- CGI,
Webware, Skunkweb, etc.  Skunkweb isn't Windows-compatible, however,
which may be a concern.  CGI is slow, but easy to install.  Webware
should install anywhere, but may take a bit more server setup -- if
possible I would recommend using Apache, but it doesn't rule out IIS. 
There's a bunch of other frameworks -- Quixote, Albatross, jonpy... I
can't remember them all, but they are all actually fairly similar in
scope.  I would *not* recommend Zope, which is a content management
application and framework, not a good web development framework.

CGI will be a little harder to program -- be sure to use cgitb if you go
that route -- but the application you describe sounds fairly
straight-forward.

For the graphics, you may want Reportlab for nice reports, or another
graphic library if you want to inline the graphs in HTML pages.  I don't
believe PIL does what you want -- there are wrappers for gnuplot, and
there's a *ton* of other graphing packages (I think the scientists tend
to use lots of graphs and lots of Python).  Search Parnasus.

  Ian