document pipelining about node-redis

[rjack]

Does this mean that commands are serialized so that

client.set("key", "value");
client.set("key", "new value");
client.get("key", redis.print);

is always guaranteed to print new_value because commands are queued and sent in order, even if the single command calls are async?

Sorry if I'm asking an obvious question, but I still have to get used to node async paradigm.

[mranney]

Yes, the commands on a single client connection are always run in order, and
the callbacks are always invoked in order.

On Fri, Apr 1, 2011 at 12:42 AM, rjack <
[email protected]>wrote:

Does this mean that commands are serialized so that

client.set("key", "value");
client.set("key", "new value");
client.get("key", redis.print);

is always guaranteed to print new_value because commands are queued and
sent in order, even if the single command calls are async?

Sorry if I'm asking an obvious question, but I still have to get used to
node async paradigm.

Reply to this email directly or view it on GitHub:
#49 (comment)

[rjack]

Thanks :)

[courtneycouch]

Took a look around the source and don't see any logic to handle pipelining, and watching network activity while using node_redis I don't see any "automatic pipelining".

Might have just missed the pipelining pieces when looking around. Any chance for some quick addition to the README with a short example of pipelining a set of commands?

[mranney]

Pipelining means sending a command without waiting for the response from the previous one. If you do this:

client.set("key1", "val1");
client.set("key2", "val2");

This will be pipelined. The second client.set command is sent before waiting for the response from the first set command .

[courtneycouch]

Ok so this is different than what pipelining means for redis I believe where
multiple commands can be sent simultaneously on the same connection. Then
results are delivered back together. Like a single batch of commands.
On Jun 12, 2011 5:30 PM, "mranney" <
[email protected]>
wrote:

[jhuttner]

Is a single TCP connection connection used in the client.set("key1"); client.set("key2") example above, Matt? If not, then I don't think node_redis is pipelining in the technical sense of the word. Seems to me like this example results in two requests to the redis server, and two separate responses - the lack of a callback means the library simply doesn't execute a callback. Is this the correct understanding?