HSET on existing field returns 0

[Jason J. W. Williams]

Hello,

I'm a little confused by the behavior of HSET.

Assuming a key named "test_hash" and fields named "foo1" and "foo2"
the following occurs:

1.) HSET test_hash foo1 test_value

This returns 1 as expected.

2.) HSET test_hash foo2 test_value2

This returns 1 as expected.

3.) HSET test_hash foo1 test_value_new

This returns 0 which to me indicates the operation failed. However...

4.) HGET test_hash fool1

Returns test_value_new.

Why does #3 return a 0 which indicates a failure?

-J

[Konstantin Merenkov]

Maybe it's not about failing but about overwriting? Like, 0 - there
was a key with this name already.
In fact HSET cannot fail.

--
Best Regards,
Konstantin Merenkov

[Jason J. W. Williams]

I would assume HSET would fail if the system hit memory exhaustion no?
Just seems confusing for 0 to be a failure condition some places and
an indication of an overwrite others.

-J

> --
> You received this message because you are subscribed to the Google Groups "Redis DB" group.
> To post to this group, send email to redi...@googlegroups.com.
> To unsubscribe from this group, send email to redis-db+u...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/redis-db?hl=en.
>
>

[Konstantin Merenkov]

On Wed, Jul 28, 2010 at 9:19 AM, Jason J. W. Williams
<jasonjw...@gmail.com> wrote:
> I would assume HSET would fail if the system hit memory exhaustion no?
> Just seems confusing for 0 to be a failure condition some places and
> an indication of an overwrite others.
>
> -J

I was right, documentation says: "If the field already exists, and the
HSET just produced an update of the value, 0 is returned, otherwise if
a new field is created 1 is returned."

Anyway, check the doc, it describes return values for all functions -
http://code.google.com/p/redis/wiki/CommandReference

[Jason J. W. Williams]

Yeah...I know what doc for HSET says. I'm disputing the behavior as inconsistent with the general doc on Integer Replies.

-J

Sent via iPhone

Is your e-mail Premiere?

[Damian Janowski]

On Wed, Jul 28, 2010 at 2:49 AM, Jason J. W. Williams
<jasonjw...@gmail.com> wrote:
> Yeah...I know what doc for HSET says. I'm disputing the behavior as inconsistent with the general doc on Integer Replies.

What's the general doc on Integer Replies?

Many commands return 0 or 1 depending or whether the key existed or
not, so I don't see the inconsistency. If the command really failed
(like you provided the wrong amount of arguments), Redis will reply
with an error (-ERR).