godaddy / node-priam Goto Github PK

Currently, if I wanted to add multiple queries to a batch I would have to call "add" multiple times. The request is for the add function to allow an array of query/batch objects to add.

In node-cassandra-cql documentation, says:

"If you are using Cassandra 2.x and you want to enable all the latest features in the Cassandra binary protocol v2 (ie: batches), you should reference version 0.5.x:"

This driver uses batches, but using node-cassandra-cql 0.4.4. There's not something wrong?

If your connectionResolver returns a property, it will overwrite the config property that was given when instantiating the driver. This may be intended, but it's preventing us from sharing a specific connectionResolver, because we can't override one of the properties (specifically keyspace). People may be relying on this behavior, so I'm not sure we can just switch it outright, but maybe we can add a feature to specify which direction the override occurs? I'd be happy to create a pull request, but I wanted to get some opinions before I started coding.

I would expect that the callback should have an error.

When the driver gives an error the callback logic in base-driver is trying to call map on a non array object (results)

I'm encountering some timeouts while querying against a C* database. In order to diagnose what's going on, I'd like to do some additional logging and timing measurements. I think this would be much easier to do if a priam instance is an EventEmitter. In fact, I started a branch to do just that, but I wanted to run this by you in case you have any better ideas.

In looking through the code for basedriver and the node-cassandra-cql driver, here are the events I determined that we could emit:

What are your thoughts about this approach? This should give consumers a lot of flexibility in how they do debugging and logging for their application.

Priam should detect these clauses and not attempt to write timestamp clauses at the DML level.

Related: https://datastax-oss.atlassian.net/browse/NODEJS-86

Batching performance improvement.

I wonder if it is possible to pass custom options to the datastax driver, an address resolution policy for example.
Or is there another procedure that I've overlooked to translate cassandra node addresses?

Do I need to type hint in param()?

Query building should support fluent-style and promises syntax.

This might be a very specific use case, but for testing small scripts (not using an express server), it might be nice to offer a closeConnection() method on the Driver so that scripts can be completed without having to ^C kill them. I was able to do this by manually calling the closePool() method, but currently the closePool method requires a "pool" argument which seems to be more for internal use (see code snippet below).

var db = require('priam')(options);

db.cql("SELECT * FROM users", [], {consistency: db.consistencyLevel.QUORUM}, function(err, users) {
    if(err) {console.log("ERROR: " + err);}
    console.log(users);
    db.closePool(db.pools.default, function() {
        console.log("Pool closed");
    });
});

I have an object that uses priam, and I want to test it withou connect to my cassandra database, how do I do this?

It says to call after execute() but I get an error. Any examples please?

I'm doing like this:

return db.beginQuery().query(q.user.getUser)
.param(email)
.options({executeAsPrepared:true})
.execute()
.single()

From the discussion #45 (comment)

indexzero: Using instanceof on custom prototypes makes working with that prototype more difficult. For example I cannot write a Facade which wraps Batch because it isn't the exact same prototype. I wouldn't go as far as to say "instanceof considered harmful" (although other folks have), but I generally only use it with prototypes native to the language.

scommisso: I can sniff whether or not it is a batch a different way. I'll address this.

Feature detection instead of instanceof seems like the best way to go.