Python is great, but why do python frameworks suck that much !?

@Dogman: I tried rails, but I don’t like it. I like to have the right to choose how I will do my stuff. I don’t like to be forced to use a specific way to do so.

@marcus: I’m currently reviewing django, and so far, I am pleased with what I see in the documentation section.
Now, I will install it and see how it goes. But on the main points I was ranting here, they are addressed in a rather nice way.

@Iongion: My day job is DBA. I work with stored procedures and design db schema all day long. In the site I’d like to port to python, performance will be a crucial key. For that, I implemented stored procedures into the master DB node that should take care of the most tedious part.
The site I’m talking about is mostly static pages, it’s the service it provides that needs to be optimized, because it might end up with many concurrent requests to the db.
An ORM is great if you want basic insert/update/delete, but when your need is past this, it’s more troubles than help.
And the syntax is terrible.

I did not looked to turbogears as a mean to shorten my development time, but as a tool to abstract myself from tedious bits of web programming.

After doing 8 years of PHP in a professional way (and before my DBA job) I can say that I think I have come to the end of what it can bring me.
As the service is already a python application (called from php with a shell call, urgh…), the portage to a python web site makes totally sense.

@tim: I’m browsing the documentation of web.py right now. Thanks for the hint.
As for the PHP+XSL, go to http://www.quixml.org .
I’ve put this together when I was happy enough of the result.

This site itself is build itself with this wrapper, and if you use firefox and you ask to see the source, you will only see the XML.

The wrapper has just a little bit of pre-processing, that allows me to plug in to merge several style sheets into 1.

In simple, I define business logic in classes, and the “public” php pages are composed like that:

< ?php
$objXml->setRendering('html');
$objXml->setXsl('org.quixml.index');
$objXml->start();
   $objXml->general();
   $objXml->open('DATAS');
      $objXml->att('demo','true');
      $objXml->data('feeling','Happy');
   $objXml->close();
$objXml->close();
?>

The xsl style sheets are served from an /xsl/{xsl path} url.
The setRendering() allows me to choose the media type. Standrd html, lo-fi, text, mini (iphone or pda)…
An infinite number of rendering set can be defined, and simply switched with this instruction.

The wrapper handle the client detection. Everything modern enough (Firefox/mozilla, ie 5.5+ and safari 1.3+) receive XML + XSL.
Everything else have the XML transformed locally on the server to be sent as simple HTML4.

Simple and effective…

It’s certainly far from perfect, but it do a good job on some toy project I had/have (but like many other, that I have not finished…)

I thought that it could maybe interest other peoples, and created that simple site.
But documentation is awful, I know.

Twitter me (I’m following you now) if something seems not clear enough.

@Eric: THANK YOU !!!!
It’s almost a replica (well, more probably on the other way) of what my PHP xsl wrapper does.

Django looks great, but again, if I can use XML + XSL for the view part, I will be quite happy.

By the way, the link to your blog seems not to be working.
I’ve got a blank page.

Thanks a lot for the links Eric. I’m gonna check them tomorrow.
I’ve just quickly read the 4suite project examples page, and I really like it.

The next step I see coming is that I don’t exactly know what is an wsgi compliant application, thus how to implement it.

Turbogears and django seems to handle that themselves, but now I’m too sleepy to understand it.
I simply will do that tomorrow.

@MV: maybe I am, well, not using the wrong tool, but at least looking at the wrong one.

But I had similar grip for every framework I ever have looked up, so I’m asking myselfif it’s not my way of working that is simply not adapted to a framework.

And by the way, it’s not because my way of working don’t fit in a framework that it makes it a basic level of programming.
Portability amongst DB is simply not something I want.
Limiting the templating by a reduced set of instruction (and possibilities) is not what I’m looking for.
Creating another blog or wiki engine in python is not what I’m looking for

I have a specific need, for a service who is already running in python, and as the performances of the front end to this service comes in question, a portage of this front-end to Python makes sense.

I don’t want to reduce the functionnalites just to add an “powered by python” at the bottom of the page.