Web.py, a python micro framework

by jacquesm • 17 years ago

21 37 17 years ago

erlanger5• 17 years ago

I'm surprised this has never been mentioned. web.py is my first choice for small ad-hoc websites these days over PHP - the Python standard lib is a huge plus and I've noticed that Python scales better than PHP on a small VPS.

The foundation of web.py itself is pretty solid, so even after you replace the ORM with SQLAlchemy, and Templetor with Mako, you have a nimble, functional framework. I wish they figured out what to do with the globals though, that make scripts feel messy.

icey• 17 years ago

It's been mentioned a bunch of times, it just falls out of the dupe filter after some amount of time.

http://searchyc.com/web.py

Web.py's author (aaronsw) posts here occasionally.

nostrademons• 17 years ago

It has been mentioned a few times - the creator (aaronsw) is a YC alum and occasionally stops by this forum.

It's fine for small ad-hoc websites, but it tends to run out of gas with larger, more complicated sites. The globals are a big issue, and you frequently end up writing wrappers around base web.py classes to fill in missing functionality. Might as well just start with Django and get the missing functionality to begin with.

erlanger5• 17 years ago

Django unfortunately forces you into a restrictive architecture.

jacquesm7• 17 years ago

I've been busy learning django/python for a few weeks now and I'm reading pretty much all day. Someone on a forum somewhere mentioned this as a solution for some of the frustration a couple of people felt while working with django.

My own personal list of bad stuff to date (about django, not web.py):

There is a lot of 'magic' going on, in other words there are lots of things that have to be done 'just so' or it will not work and you'll get a totally meaningless error message (in fact the error messages are some of the worst I've ever seen).

Simple things (like extending the user model) are done pretty kludgy and not standardized as part of the framework, which I really think they should be. It would be easy enough to 'augment' the user table in a documented manner for instance.

Pretty much all of the documentation seems to teach you how to set up your applications in a non-portable way, in other words the application has to live in the named directory that it was conceived in.

I'm definitely going to play around with webpy under wsgi as soon as I have a little bit of time.

erlanger5• 17 years ago

That's the problem with most frameworks. The Auth looks nice but is a pain to implement just right...Profiles are painful enough in Django. I've noticed that in the end you usually have to rewrite the core components, so it's better to choose software that anticipates this. Django is great for run-of-the-mill sites, just like Rails -- I love generic views. I'm more likely to use Drupal for that class of website though.

apgwoz3• 17 years ago

I tried using web.py on multiple occasions and found that it never quite fit the bill. I'd love a framework that is in between this and django, but I'm afraid that django is just probably the way to go regardless... There's too many benefits of the (almost) reusable apps, and of course the admin for quick data prototyping, and sometimes production admin console.

tjpick• 17 years ago

pylons?

apgwoz3• 17 years ago

I knew someone would suggest pylons. I haven't really checked out the community around pylons, but my guess is that there isn't quite as much reusable stuff out there as there is for django. I should look again though.

moe• 17 years ago

Werkzeug

anatoly2• 17 years ago

I like web.py's approach to templates. A few simple uses of $, if I remember correctly. A minimalist aesthetic. A typical Python framework's templating system is contrived and looks ugly by comparison.

pssdbt2• 17 years ago

Django's template system is the sexiest I've seen (mostly because of the simple syntax and inheritance). I've been meaning to check out web.py for a while now, will have to see how web.py templating compares.

anatoly2• 17 years ago

Sorry; when I see those

  {% %} {{ }}

I feel nauseated.

Also, it's bizarre to use a language (Python) that uses indentation instead of endif's, closing braces, etc., and then turn around and require endif's and endfor's in your templating system. web.py's templates get that right: just use the Python style.

erlanger5• 17 years ago

Mako's the best I've found. Templetor looks nice but is awful because (of all things) whitespace is significant! That's a no-go for a template system.

apgwoz3• 17 years ago

The best I've ever seen (though I haven't seen an implementation for Python) is straight up x-expessions (sexps) with unquoting and unquote-splicing.

mr_luc5• 17 years ago

Did something new ("Hacker News") happen with web.py?

...

Ah, no. Hmm. Well, I've proposed this before(http://news.ycombinator.com/item?id=750998), but what do you guys think about the adoption of this rule:

"If you're posting to inform us of the existence of something, assume we already know."

Details and examples in the post linked above.

Assume that we know about web.py, especially since its creator is from the first set of YC startups.

jacquesm7• 17 years ago

I figured if the url was posted before the dupe filter would catch it, apologies.

Also, and this is something that keeps coming up here, we did not all sign up to HN on the same day, so we do not all know what has gone before.

Lastly, what is not useful to you may be useful to someone else, I think that if people here would apply your "If you're posting to inform us of the existence of something, assume we already know." rule that a lot of very interesting content would not make the cut.

mr_luc5• 17 years ago

[ I should preface this all with a giant In My Humble Opinion. I think that this rule has the potential to be a hack that solves a problem, so I'm arguing for it -- it's not a rule that's been adopted, just something that seems promising to me. This should in no way be construed as criticism of the poster, particularly since web.py is very hackerly. ]

Your objection is not only valid, but correct! "A lot of very interesting content would not make the cut."

I think that's a feature, not a bug.

That's the point of HN! It's Hacker News, not Hacker Stuff.

For instance: did someone write an interesting article talking about their experience with actually using web.py, in a way that is likely to be benefit other python hackers?

Then that would be news, and it would qualify! And much more so if they were your own experiences.

But to inform us of the existence of things that are not new -- no offense, but we could all just noodle around in Wikipedia and post interesting, hackerly content until the sun went cold.

Luyt• 17 years ago

One step up would be CherryPy.

blasdel2• 17 years ago

I'm using CherryPy for a project right now, and it certainly has some weirdness to it -- it tries to abstract HTTP just a little bit too much, and ends up doing some seriously stupid shit:

By default all exposed methods can be called as GET or POST, with keyword arguments found appropriately.

cherrypy.engine keeps a bunch of static state internally -- once you've started it, you can't restart it (or even catch an exception in start() and retry) with different settings -- you have to start over again in a new process (as you can't unload modules in cpython).

abdels• 17 years ago

Web.py reminds me greatly of Sinatra http://www.sinatrarb.com/, a minimalistic very agile ruby web framework.

These sort of frameworks are ideal for developing APIs or stand alone (RESTFUL) services, anything that does not rely heavily on forms or a proper UI experience. Sinatra for example can only have one controller, so any app you develop will naturally do one thing (and do it well).

They're also perfect for prototyping ideas!

blasdel2• 17 years ago

Google AppEngine's 'webapp' is web.py done right.

Web.py, a python micro framework

Related Stories