Re: [flask] Two extensions released
- From:
- LeafStorm
- Date:
- 2010-06-13 @ 06:01
I don't have very much experience with MongoDB, but from what I
understand its databases use completely different organization (Mongo
has databases containing collections of documents, whereas Couch uses
databases which just have documents) and queries (Mongo uses SQL-like
finds and indexes handled entirely at the application level, while Couch
uses MapReduce-style views stored in the database) than CouchDB, so code
would not really be portable between the two.
Most of the Flask-CouchDB code is based on features present in the
couchdb-python library that do not appear to be present in PyMongo, like
the Document classes and view synchronization. Really, forking would
leave a lot of extra, non-working code that you would have to get rid
of, so it would probably be more efficient to start your own
Flask-MongoDB (or whatever) project from scratch. You could find some
useful insights in the code, however.
PS: I think the major design difference between Couch and Mongo is that
Couch is experimenting with new paradigms designed specifically for
document-oriented databases, while Mongo is using some of the cool stuff
from DODBs but keep enough SQL-ish features to keep relational DB
experts happy. IMHO, I like Couch's approach better because it lets you
do some really cool stuff no one had really thought of before, and
really MapReduce is a lot better fit for DODBs than SQL-type queries.
On 06/12/2010 03:31 PM, Thadeus Burgess wrote:
> Leaf, I might fork and modify for mongodb
>
> --
> Thadeus
--
Regards,
LeafStorm (http://www.leafstorm.us/)
Re: [flask] Two extensions released
- From:
- Armin Ronacher
- Date:
- 2010-06-13 @ 10:18
Hi,
On 2010-06-13 8:01 AM, LeafStorm wrote:
> [...] IMHO, I like Couch's approach better because it lets you
> do some really cool stuff no one had really thought of before, and
> really MapReduce is a lot better fit for DODBs than SQL-type queries.
The best feature of Couch is still that it speaks HTTP. You can hook a
cache between your database and the application which is awesome.
Regards,
ARmin