Django Vs Rails

[flamesrock]

Firstly, this topic is NOT intended for trolling or starting any flame
wars.

I want to know if anyone has experience with these frameworks, and if
so, how do they compare? Which one do you prefer?

Going back to school in a few days, I simply don't have the time to try
both.

-thanks
http://djangoproject.com
http://rubyonrails.com

 

[D H]

Firstly, this topic is NOT intended for trolling or starting any flame
wars.

Whatever you say, "flamesrock".

 

[gene tani]

http://griddlenoise.blogspot.com/2005/07/python-off-rails.html
http://www.boddie.org.uk/python/web_frameworks.html

 

[flamesrock]

Thanks for the links, gene tani.

D H,
'flamesrock' refers to the Calgary Flames, not the act of flaming.

 

[Diez B. Roggisch]

http://griddlenoise.blogspot.com/2005/07/python-off-rails.html

I don't think that opinion is very founded - especially what he thinks
metaprogramming is about, and what not.

What he seems not to grasp is the crucial difference between django and
rails/subway (at least in the way the yare _used_ by his examples):

- rails/subway reflect over a existing table. They create OR-mappings
based on that. You only specify exceptional attributes for these mappings.

- django specifies the whole meta-model in python - and _generates_
the SQL/DDL to populate the DB. So obviously you have to be more verbose
- otherwiese you won't have the fields you need.


I personally prefer the latter approach - I want my tables modeled after
my objects, not the other way round. I tried to find out if subway and
rails can do the same - that is, generate the sql. For subway the lack
of documentation prevented that, and I didn't find it in rails , too.

And there is at least one shortcoming to the first approach, when using
the most popular RDBMS, MySQL: The lack of foreign key constraints makes
me wonder how to automatically infer 1:n or m:n relationships. From a
rails tutorial, I see that one has to declare these too:

http://www.onlamp.com/pub/a/onlamp/2005/01/20/rails.html?page=5


But maybe someone who has expirience with subway or rails can elaborate
on this?

Diez

 

[gene tani]

Here's another, (i'm just flipping thru del.icio.us and furl tags, not
endorsing any viewpoints. I've never looked at Django or Subway, but I
do need to look at zope 3, i think):

http://www.magpiebrain.com/archives/2005/08/14/rails_and_django

 

[Jaroslaw Zabiello]

Dnia 5 Sep 2005 19:06:51 -0700, flamesrock napisa³(a):

Firstly, this topic is NOT intended for trolling or starting any flame
wars.

I want to know if anyone has experience with these frameworks, and if
so, how do they compare? Which one do you prefer?

Django's ORM does not work with SQL Server (only Postgresql, MySQL and
SQlite). it would be problem to use Django for m$ based intranets. Rails
has more features, better docs and seems to be more mature than Django. And
last but not least, Django has no official release up today.

 

[D H]

D H,
'flamesrock' refers to the Calgary Flames, not the act of flaming.

It was just a joke about your statement and your name. I thought it was
obvious enough that a smiley wasn't necessary. I don't care though,
flames happen on comp.lang.python all the time.
Go with Rails. Django is only like a month old. Unless you are more
comfortable using python than ruby.

 

[bruno modulix]

D H wrote:

(snip)

Go with Rails. Django is only like a month old.

Please take time to read the project's page. Django has in fact three
years of existence and is already used on production websites, so it's
far from pre-alpha/planning stage.

 

[D H]

D H wrote:

(snip)



Please take time to read the project's page. Django has in fact three
years of existence and is already used on production websites, so it's
far from pre-alpha/planning stage.

Don't make any assumptions about what I have and haven't read. Django
was only publicly released in mid-July. Less than two months ago.
I never implied it was "pre-alpha" or in a "planning" stage.
The Ruby framework is also more mature and much much more tested than
Django. But even then I am not stating one is inherently better than
the other. Both seem very well designed. But if you have no clue about
either project like the original poster, then I'd recommend Rails,
unless you are already comfortable with Python (and not used to Ruby),
which case go with Django.

 

[Alan Kennedy]

[D H]
[bruno modulix]

Please take time to read the project's page. Django has in fact three
years of existence and is already used on production websites, so it's
far from pre-alpha/planning stage.

But the APIs still aren't 100% stable.

http://www.djangoproject.com/screencasts/model_syntax_change/

 

[bruno modulix]

Don't make any assumptions about what I have and haven't read.

Don't make the assumption that the advice to read the project's page was
directed to you only !-)

Django
was only publicly released in mid-July. Less than two months ago.
I never implied it was "pre-alpha" or in a "planning" stage.

That what I understood from your post, and I guess most readers would
understand it that way.

Now the fact that Django, while being young as a publicly realeased
project, has in fact 3 years of existence *on production websites* is
IMHO worth mentioning, so peoples who haven't read the project's page
yet don't make the assumption that it must be too young to be usable in
production. The "take time to read..." advice was directed to these
people as well !-)

Regards

 

[gene tani]

just to make it really easy, there's some really good blogs and
detailed analyses of the frameworks, competition is good:

http://del.icio.us/tag/rails+django

 

[Terry Reedy]

D H wrote:

(snip)

Please take time to read the project's page. Django has in fact three
years of existence and is already used on production websites, so it's
far from pre-alpha/planning stage.

One can also check it out in action.
http://www.ljworld.com/ is the site it was developed for.

 

[Greg McIntyre]

I tried to find out if subway and
rails can do the same - that is, generate the sql. For subway the lack
of documentation prevented that, and I didn't find it in rails , too.

In Rails you can do that with the command:

$ rake db_structure_dump

However I think it's not the prescribed way of using it because it
tends to involve losing all your data every time you make a schema
change. I think they recommend doing this once at the start of
development if you don't have a DB schema yet. Once you're up and
running, if you rely on ALTER TABLE type commands and manually update
your code to match the new schema, although it's more work it leaves
your data intact. I think that's the idea anyway.

And there is at least one shortcoming to the first approach, when using
the most popular RDBMS, MySQL: The lack of foreign key constraints makes
me wonder how to automatically infer 1:n or m:n relationships. From a
rails tutorial, I see that one has to declare these too:

http://www.onlamp.com/pub/a/onlamp/2005/01/20/rails.html?page=5


But maybe someone who has expirience with subway or rails can elaborate
on this?

From my experience with Rails, the OR mapping isn't hugely automated.

That is, some manual work is required if you change the schema, to
update the code that operates upon it. However although this is another
step it is pretty trivial due to the metaprogramming style methods.
Besides, if you're altering the schema you tend to have to update some
code anyway as you're most likely altering the functionality of the
system a little. And as I said above, a loose/manual OR mapping has its
benefits; I don't feel anxious about losing my data/schema when using
Rails because I know it leaves it alone.

 

[flamesrock]

Wow- thanks for all of the replies. I'm torn.

On the one hand, I'm fluent in Python and love it.

On the other, Rails seems to have a brighter future, and is a bit more
featureful (at this time.) However the only Ruby I know is what I've
already learnt with Python(even though I would like to learn it).

How difficult would it be to learn Ruby+Rails, assuming that someone is
already skilled with Python?

Is it worth it?

 

[James]

I actually like the framework to reflect on my database. I am more of a
visual person. I have tools for all my favorite databases that allow me
to get a glance of ER diagrams and I would rather develop my data
models in these tools rather than in code. Further more I rather like
the idea of parsimonious use of code (which is probably why I use
Python in the first place) and do not really like manually specifying
data schemas in code as much as possible.

Is some familiar with a Python Framework that builds by reflection.

 

[Marek Kubica]

Hello!

On the other, Rails seems to have a brighter future,

Why that? Django is not yet released and everybody is talking about it.
Like it happened with RoR.

How difficult would it be to learn Ruby+Rails, assuming that someone is
already skilled with Python?

Learning Ruby: quite trivial, as Ruby is like Python, but sometimes there a
Ruby-way and a non-Ruby-way (codeblocks and stuff like this) and to write
nice Ruby programs you better write in the Ruby-way.

Is it worth it?

Well, I've learned it, because it was easy but I haven't yet found a really
significant difference that makes Ruby much better than Python. You can
write some things in an very elegant way, but sometimes the Python solution
is more readable. But the docs are sometimes.. say: very compact .
As Ruby is not that hard to learn you could give it a try - maybe you'll
like it, maybe not.

RoR is not the only framework, some folks prefer Nitro.

greets,
Marek

 

[Ian Bicking]

- rails/subway reflect over a existing table. They create OR-mappings
based on that. You only specify exceptional attributes for these mappings.

- django specifies the whole meta-model in python - and _generates_
the SQL/DDL to populate the DB. So obviously you have to be more verbose
- otherwiese you won't have the fields you need.

Subway is built on SQLObject, and SQLObject can both create tables and
read schemas from an existing database. I don't know what style is
preferred at the moment in Subway, but personally I always create my
tables with SQLObject because I always end up enumerating all the
features of the tables eventually anyway.

 

[Jeff Shell]

My opinion, as posted there, was pretty immediate and only going off of
surface values. I just saw in Django what I had seen too much of in my
own code. I've written similar things in Zope and Formulator that did
all sorts of fancy automatic 'admin screen' generation, DBMS CRUD
statements. I even applied those same tricks to LDAP with some success.
So some of what I saw with Django reminded me of code that I both loved
and hated - it made my data management work easy, but it was ugly and
hard to navigate and maintain, and I could never remember even my own
shortcuts and options of what to have in my nested dictionaries and
lists and tuples that defined all of these structures.

It was also an all-too-quick response in that I perceived, wrongly,
that Django was another Rails clone. I was quickly corrected on this,
and believe I posted a correction later. As I've said in some other
posts on my site - I don't really care much for the Python "Web
Frameworks" which go off and try to mimic the technology of the day.
I've been through competing with Perl, with Servlets, with J2EE, with
PHP/ASP, and now with Rails, and just about every implementation I've
seen of these has been underwhelming, to say the least. I myself am
guilty of writing an early WebObjects inspired toolkit back in '96.
Django is its own thing, and I give it credit for that. Some recent
changes to the Django model syntax make it more palatable to my tastes,
but I doubt that I'll personally be leaving Zope 2 or 3 for any of
these systems any time soon. Zope / Principia / Bobo have all served me
well for nearly a decade now, and Zope 3 makes ZODB based Zope
development so easy and powerful that I hope to never have need of an
ORM system again. These ORM tools like SQLObject and ActiveRecord still
seem best suited for writing applications from scratch. Ian Bicking and
Django people both have stated here that you can "conveniently generate
SQL/DDL to populate the DB". If that's the situation, an object
database like the ZODB or Durus may work just as well, without the
translation to SQL and back. ORM only gets interesting to me these days
when I have to map to to large scale pre-existing systems, and even
there I'm having better experiences with other styles of storage to
application data translation, validation, security, and management.