What ORM do you use with Tornado?
Joshua Partogi
Just a quick question, with a large number of ORM in Python, what ORM
do you use with Tornado? What is the advantage of using it with
Tornado?
Thank you in advance,
Emre Yılmaz
mattd
syntax, and under very active development.
Matt Ferguson
you're looking for a true ORM, check out SQLAlchemy (and possibly
Elixir). Or join the NoSQL revolution and give MongoDB or CouchDB
(or one of the many other solutions) a shot.
Sent from my iPhone
On Jan 25, 2010, at 7:28 PM, Joshua Partogi <joshua....@gmail.com>
wrote:
Marinho Brandao
Since I started to use Python I see a problem with all "ORM"
solutions, and the problem is simple: all of them are just ORM.
What I mean is that, if you want to use SQLObject, SQLAlchemy or
Django's ORM to persist objects in a NoSQL database, it's not
possible, and the solution far enough to create a new library to do
it.
They just don't fit to the paradigm out of relational, what includes
NoSQL databases, memory caching, simple files, or every kind of data
abstraction (until a plugin here is a little harder to do) out of
relational paradigm.
When I developed with Delphi and PHP, I had my own persistency layer
libraries I made during the years and they worked very well. Instead
use the name "ORM" I prefere "Object Persistency Framework", what
supports kind of a DAO layer, but with object/relational mapping
functions.
If I had time enough I would create a new persistence abstraction to
do it, but while this doesn't happens, I use Django's ORM to
relational databases and manually persistency when no-relational. In
Tornado, I probably will start something like a hibrid of Django's ORM
and NoSQL, I'm not sure.
That's all :)
2010/1/26 Emre Yılmaz <yilma...@gmail.com>:
--
Marinho Brandão (José Mário)
http://marinhobrandao.com/
David P. Novakovic
Anyway, the question was about ORMs not object persistance libraries. There are lots of ORMs for python, I'd only ever use django orm or sql alchemy. Though I've heard storm is quite good too. :)
With anything like this, you should try them and use the one you like the most.
The only improvement I could really wish for in tornado is async db access. :)
Andrew Gwozdziewycz
<davidno...@gmail.com> wrote:
> Actually. Django ORM has been designed to work with schemaless databases and
> there will be support for various libraries released soon. Sounds like that
> would be a good framework to hook your schemaless efforts into.
Well, technically it wasn't designed for schemaless solutions. All the
solutions for doing schemaless stuff have been hacks which monkeypatch
the ORM to do what is needed. There is an effort however to make this
stuff non-hacky and officially supported.
Joshua Partogi
fit. So, is a non-blocking db server? Because after giving it a
thought, tornado would be perfect with backend that is also a non-
blocking server.
Regards.
On Jan 26, 2:38 pm, Matt Ferguson <mbf...@gmail.com> wrote:
> Tornado comes with a lightweight and simple MySQL wrapper, but if
> you're looking for a true ORM, check out SQLAlchemy (and possibly
> Elixir). Or join the NoSQL revolution and give MongoDB or CouchDB
> (or one of the many other solutions) a shot.
>
> Sent from my iPhone
>
> On Jan 25, 2010, at 7:28 PM, Joshua Partogi <joshua.part...@gmail.com>
> wrote:
>
> > Hi all,
>
> > Just a quick question, with a large number of ORM in Python, what ORM
> > do you use with Tornado? What is the advantage of using it with
> > Tornado?
>
> > Thank you in advance,
Andrew Gwozdziewycz
<joshua....@gmail.com> wrote:
> I have thought about using mongo with tornado. It seems as a perfect
> fit. So, is a non-blocking db server? Because after giving it a
> thought, tornado would be perfect with backend that is also a non-
> blocking server.
>
> Regards.
The mongo driver doesn't do async. However, there is apparently
experimental support for querying it via HTTP:
http://www.mongodb.org/display/DOCS/Http+Interface, which as you know
has an async client already in tornado.
> On Jan 26, 2:38 pm, Matt Ferguson <mbf...@gmail.com> wrote:
>> Tornado comes with a lightweight and simple MySQL wrapper, but if
>> you're looking for a true ORM, check out SQLAlchemy (and possibly
>> Elixir). Or join the NoSQL revolution and give MongoDB or CouchDB
>> (or one of the many other solutions) a shot.
>>
>> Sent from my iPhone
>>
>> On Jan 25, 2010, at 7:28 PM, Joshua Partogi <joshua.part...@gmail.com>
>> wrote:
>>
>> > Hi all,
>>
>> > Just a quick question, with a large number of ORM in Python, what ORM
>> > do you use with Tornado? What is the advantage of using it with
>> > Tornado?
>>
>> > Thank you in advance,
>
> --
> http://twitter.com/scrum8
>
Milan Cermak
On Jan 26, 12:14 pm, Joshua Partogi <joshua.part...@gmail.com> wrote:
> I have thought about using mongo with tornado. It seems as a perfect
> fit. So, is a non-blocking db server? Because after giving it a
> thought, tornado would be perfect with backend that is also a non-
> blocking server.
>
There is a project for asynchronous MongoDB library on GitHub
http://github.com/fiorix/mongo-async-python-driver However it uses
Twisted and is not production ready yet.
I too would love to see a Tornado enabled MongoDB library.
Marinho Brandao
> Actually. Django ORM has been designed to work with schemaless databases and
> there will be support for various libraries released soon. Sounds like that
> would be a good framework to hook your schemaless efforts into.
nope, as Andrew said, it did'nt. It just don't fit
> Anyway, the question was about ORMs not object persistance libraries. There
it's 2010, and we are talking about the web and the cloud. No matters
the topic is "ORM", the truth is that caching data and distributed
information is an important part of this topic.
even if is not the best thing (the coupling), ORMs are mostly
persistence libraries too, and they just don't fit to a growing and
important part of this topic.
> are lots of ORMs for python, I'd only ever use django orm or sql alchemy.
> Though I've heard storm is quite good too. :)
ha, yeah, I forgot about Storm :)
> With anything like this, you should try them and use the one you like the
> most.
when I tried Storm and others I don't remember their names (when I
began in Python 5 years ago), there were no choice about what I'm
talking. Maybe today the things are differents on some of them.
BTW, the DB API also does not really fit to NoSQL paradigm, but
whatever, I think this is not "the" problem, since the persistence
layer should be the responsible about that.
> The only improvement I could really wish for in tornado is async db access.
> :)
yeah, also :)
> Well, technically it wasn't designed for schemaless solutions. All the
> solutions for doing schemaless stuff have been hacks which monkeypatch
> the ORM to do what is needed. There is an effort however to make this
> stuff non-hacky and officially supported.
BTW, it's a BIG job (almost to make a new one, keeping the API definition)
best regards,
Joshua Partogi
On Jan 26, 10:20 pm, Andrew Gwozdziewycz <apg...@gmail.com> wrote:
> On Tue, Jan 26, 2010 at 6:14 AM, Joshua Partogi
>