Re: [Pharo-users] Serializing to/from Json
On Aug 29, 2011, at 9:58 PM, Francois Stephany wrote:
On 29/08/11 21:27, Larry White wrote:
Hi,
I'm hoping to use CouchDB for storage on a seaside application. It would be lovely if I didn't have to serialize/de-serialize by hand to json for all my persistent objects, and could use instead something like a smalltalk version of Google's Gson library. Is there anything like that?
I've tried the JSON library bundled with SCouchDB and it seems to automatically handle simple objects (eg, where the inst vars reference Strings or Numbers), but not handle more complex objects. Also, I haven't seen anything that de-serializes from json.
Does it need to be JSON? There is a new binary serializer: http://rmod.lille.inria.fr/web/pier/software/Fuel Marcus -- Marcus Denker -- http://marcusdenker.de
On Tue, Aug 30, 2011 at 10:23 AM, Marcus Denker <marcus.denker@inria.fr> wrote:
Does it need to be JSON?
Since CouchDB stores data in json - I guess the answer is very likely yes, since just storing serialized blobs inside CouchDB would loose all querying and indexing available. In that case some key-value store might be a better choice than CouchDB Davorin Rusevljan http://www.cloud208.com/
On Tue, Aug 30, 2011 at 9:40 AM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 10:23 AM, Marcus Denker <marcus.denker@inria.fr> wrote:
Does it need to be JSON?
Since CouchDB stores data in json - I guess the answer is very likely yes, since just storing serialized blobs inside CouchDB would loose all querying and indexing available.
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.<http://www.cloud208.com/> -- Mariano http://marianopeck.wordpress.com
BTW....I did experiments between Fuel and Riak, maybe that helps in something: http://marianopeck.wordpress.com/2011/08/16/when-fuel-meets-riak-nosql-datab... On Tue, Aug 30, 2011 at 3:27 PM, Mariano Martinez Peck < marianopeck@gmail.com> wrote:
On Tue, Aug 30, 2011 at 9:40 AM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 10:23 AM, Marcus Denker <marcus.denker@inria.fr> wrote:
Does it need to be JSON?
Since CouchDB stores data in json - I guess the answer is very likely yes, since just storing serialized blobs inside CouchDB would loose all querying and indexing available.
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.<http://www.cloud208.com/>
-- Mariano http://marianopeck.wordpress.com
-- Mariano http://marianopeck.wordpress.com
On 30 Aug 2011, at 16:27, Mariano Martinez Peck wrote:
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.
It could be (not knowing the specific database) that the db understands the contents and can query on the properties, which would make huge differences in performance. Say you store: 1 -> { "id" : 1, "name" : "one" } 2 -> { "id" : 2, "name" : "two" } ⦠99999 -> ⦠You could then ask the db for the object where name = "one hundred" without reading all of them (but the db would have to do that of course). Maybe even indexes are possible. Sven
Couch stores data as Json. Its query capabilities are based on json pair names. I'm just trying it out but I believe that, for example, if a User object was a blob, you couldn't issue a query like "fetch user with username = 'foo'". You would have to read all the users into Smalltalk and do a select: Anyway that's my impression of how Couch works. On Tue, Aug 30, 2011 at 10:27 AM, Mariano Martinez Peck < marianopeck@gmail.com> wrote:
On Tue, Aug 30, 2011 at 9:40 AM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 10:23 AM, Marcus Denker <marcus.denker@inria.fr> wrote:
Does it need to be JSON?
Since CouchDB stores data in json - I guess the answer is very likely yes, since just storing serialized blobs inside CouchDB would loose all querying and indexing available.
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.<http://www.cloud208.com/>
-- Mariano http://marianopeck.wordpress.com
On Tue, Aug 30, 2011 at 4:27 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.
let's say you have a class: Object subclass: #MyClass instanceVariableNames: 'a b' classVariableNames: '' poolDictionaries: '' category: 'Example' and you serialize it into the json to something like { '_id': 'lkjjfdjdjdj', 'a': 'value of a', 'b': 'value of b' } than you can ask CouchDB to fetch you all object where for instance a is 'Aha!' while if you serialize it as: { '_id': 'lkjjfdjdjdj', 'fuel_blob': 'AAXsdfkljijS3SM .....' } CouchDB has no idea how to look into the a, and select appropriate objects. Davorin Rusevljan http://www.cloud208.com/
On Tue, Aug 30, 2011 at 3:40 PM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 4:27 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.
let's say you have a class:
Object subclass: #MyClass instanceVariableNames: 'a b' classVariableNames: '' poolDictionaries: '' category: 'Example'
and you serialize it into the json to something like
{ '_id': 'lkjjfdjdjdj', 'a': 'value of a', 'b': 'value of b' }
than you can ask CouchDB to fetch you all object where for instance a is 'Aha!'
while if you serialize it as:
{ '_id': 'lkjjfdjdjdj', 'fuel_blob': 'AAXsdfkljijS3SM .....' }
CouchDB has no idea how to look into the a, and select appropriate objects.
Interesting. Now I understand. So I guess the same happens with all the rest of NoSQL databases? I am right no assume that if I want to be able to do queries without bringing the whole graph into memory, I need to store the data in JSON ? thanks! -- Mariano http://marianopeck.wordpress.com
On Tue, Aug 30, 2011 at 4:43 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
Interesting. Now I understand. So I guess the same happens with all the rest of NoSQL databases? I am right no assume that if I want to be able to do queries without bringing the whole graph into memory, I need to store the data in JSON ?
Not necessarily, CouchDB and MongoDB are based on the json, but some are more close to pure key value stores, where value is usually just some blob and db does not care much about. In combination with such databases Fuel is a great fit. I _think_ TokyTyran is such db, and memcached. Davorin
Well, you can also store binary contents in CouchDB [1]. You can attach zero or more binary attachments to a Json document in CouchDB. You could use this in combination with Fuel. Add the properties you want to search for to the Json document and leave all the other data in binary format. Jan. [1] http://guide.couchdb.org/editions/1/en/api.html#attachments On Tue, Aug 30, 2011 at 4:43 PM, Mariano Martinez Peck < marianopeck@gmail.com> wrote:
On Tue, Aug 30, 2011 at 3:40 PM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 4:27 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
I don't understand. What is the difference from the querying and indexing point of view between serializing with JSON or a binary serializer.
let's say you have a class:
Object subclass: #MyClass instanceVariableNames: 'a b' classVariableNames: '' poolDictionaries: '' category: 'Example'
and you serialize it into the json to something like
{ '_id': 'lkjjfdjdjdj', 'a': 'value of a', 'b': 'value of b' }
than you can ask CouchDB to fetch you all object where for instance a is 'Aha!'
while if you serialize it as:
{ '_id': 'lkjjfdjdjdj', 'fuel_blob': 'AAXsdfkljijS3SM .....' }
CouchDB has no idea how to look into the a, and select appropriate objects.
Interesting. Now I understand. So I guess the same happens with all the rest of NoSQL databases? I am right no assume that if I want to be able to do queries without bringing the whole graph into memory, I need to store the data in JSON ?
thanks!
-- Mariano http://marianopeck.wordpress.com
On Tue, Aug 30, 2011 at 4:43 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
I am right no assume that if I want to be able to do queries without bringing the whole graph into memory, I need to store the data in JSON ?
and some other issues might be connected. For instance, I think you could persuade CouchDB to do replication of only some records based on the json data. Davorin
Ok. I see. Thanks a lot guys for the info, I am learning a lot :) On Tue, Aug 30, 2011 at 3:54 PM, Davorin Rusevljan < davorin.rusevljan@gmail.com> wrote:
On Tue, Aug 30, 2011 at 4:43 PM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
I am right no assume that if I want to be able to do queries without bringing the whole graph into memory, I need to store the data in JSON ?
and some other issues might be connected. For instance, I think you could persuade CouchDB to do replication of only some records based on the json data.
Davorin
-- Mariano http://marianopeck.wordpress.com
participants (6)
-
Davorin Rusevljan -
Jan van de Sandt -
Larry White -
Marcus Denker -
Mariano Martinez Peck -
Sven Van Caekenberghe