Web development with Python 3.1

[Bruno Desthuilliers]

Dotan Cohen a écrit :
>> Perhaps this might better answer your questions:
>> http://docs.djangoproject.com/en/dev/topics/templates/#id1
>>
> 
> Actually, that example just proves my point!

Which one ?

> Look at this "templating code":
> {% for entry in blog_entries %}
>     <h2>{{ entry.title }}</h2>
>     <p>{{ entry.body }}</p>
> {% endfor %}

What's the problem ? Simple, clear and obvious.

> Why would I want to learn that when Python already has a real for
> loop ?

If you know even 101 Python, understanding the above code shouldn't be 
such a great challenge !-)

> I know HTML, and I have a goal of learning Python for it's
> reusability (desktop apps, for instance).

Using templates instead of harcoding HTML in the applicative code 
actually plays a big part when it comes to reusability. Telling Django's 
template loader where to look for templates is just a matter of 
configuration, so using templates you can segment your whole project in 
small, possibly reusable apps - then in another project where you need 
the same features but with a totally different presentation, it's just a 
matter of writing project-specific templates. Quite a few django apps 
work that way, and they really save you a LOT of time. This wouldn't be 
possible using your beloved antipattern...

> I don't want to learn some
> "templating language" that duplicates what Python already has built
> in!

Then use Mako - it uses plain Python to manage the presentation logic. 
And if you go for Mako, then you might as well switch to Pylons. Great 
framework too (better than Django on some parts FWIW), but you'll 
probably need much more time to be fully productive with it (power and 
flexibility come with a price...).

Now wrt/ "why having a distinct templating language", there are pros and 
cons. On the pros side, having a language that is clearly restricted to 
presentation logic prevents you (and anyone else working on the project 
- and sometimes you have to deal with well-below-average guys in your 
team) to put anything else than presentation logic in the templates.

Bless me for I have sinned - sometimes, when you're tired and under 
pressure to deliver in time, it's tempting to add a Q&D hack in the 
presentation part instead of DoingTheRightThing(tm). This can save you 
some "precious" minutes now. The problem is that usually your boss won't 
give you the resources to clean up this mess once you delivered, and 
next time you have to work on this project - fix or evolution - you have 
to deal with this hack. Then with another, yet another, until your code 
is nothing more than an unmaintainable heap of junk. Been here, done 
that, bought the T-shirt :(

Also remember that most of the time, a web app is a team effort, and 
even if *you* don't fail to temptation, others may do so. And finally, 
the team usually involve "HTML coders" - who, despite what the name 
seems to imply, are usually skilled enough to handle the presentation 
logic by themselves, so you the application programmer can concentrate 
on applicative code. From experience, they usually prefer a simple 
straightforward - even if somehow restricted - templating language.

Now, as far as I'm concerned, having Mako instead of Django's templates 
wouldn't bother me at all. But Django has it's own template system, 
which from experience get the job done (believe me, there are way worse 
alternatives), and the overall qualities and features of the whole 
framework are IMHO enough to justify learning it's templating language.

> I think that I will use the HTTP-abstraction features of Django, but
> disregard the templates. My code might be uglier, but the knowledge
> will be transferable to other projects. My ultimate goal is not to
> make the latest greatest website. My ultimate goal is to port my
> perfectly functional website to a language that will benefit me by
> knowing it.

Given the restricted and rather unintersting nature of pure presentation 
logic, you won't learn much from this part of the project anyway. You'd 
probably learn way more using Django's templates and learning how to 
write your own template tags. Or if you prefer Mako / Pylons, learning 
to make proper use of Mako's advanced features. But one thing is clear - 
if you persist on applying your favorite antipattern, you'll just waste 
more of your time. Your choice, as usual...

My 2 cents...