On Dec 3, 10:21 pm, Gerhard Häring <[EMAIL PROTECTED]> wrote: > azrael wrote: > > It logical that it would be more efficient and logical to use a object > > oriented database, but in this case I ask because of the portable > > nature of sqlite. > > > so, if I get it right, this should be possible [...] > > Did you try it? Did it work? If so,it was pure luck. Attached is a > script that shows how to do it right. > > -- Gerhard > > sqlite_serialize.py > 1KViewDownload
Hi, it is also possible to use Google Protobuf ( http://code.google.com/p/protobuf/ ) instead of pickle or json - it's portable to other languages too - you define what's the data you really wan't - it's fast, and compact I've done it on one of my projects ( see the code here http://code.google.com/p/pyshow/source/browse/trunk/src/pyshow/data/msgdb.py ), and using BerkeleyDB instead of sqlite( but that's not the point here) What I like with protobuf is that writing the proto file defines the real data structure, and that's a guaranty over time/processes/coders. I don't recommand using pickle to serialise object in a database. Because, it's easy at the beginning, but later, when you are going to 'upgrade' your objects (adding some new fields for instance, or splitting a field into a new object), you'll get a lot of troubles ( keeping the legacy object to live with the new one, implies renaming all objects ? etc.). That's what protobuf is for. BTW, I'm seeking help to start a new project that would connect protobuf to Database, like the piece of code I've done with berkeleyDB, but at a bigger scale. If anyone is interested. -- http://mail.python.org/mailman/listinfo/python-list