Sorry for confusion, I should have lined my requirements better in the
first place. Let me try to summarize:
- I can use list<blob> and query against it using secondary indexes and by
encoding my data on the client side. However, *this only allows exact
matches, not greater/lesser than *for numbers at least (not sure I need to,
but maybe). Please correct me if I got it wrong? I'm not very familiar with
playing with binary.
- My supported list of types is very limited, indeed, and the order doesn't
matter, so I could use separate list for each type. However, that makes
playing with data somewhat cumbersome and I need to have multiple clauses
in queries then, for each type.
- I could use user defined types, but I would still have to define separate
field for each value and queries would again be cumbersome.
Let's forget about "dynamic schema" as I'm a Cassandra newbie and
definitively need to study more before opening that chest of wonders.
Thanks for correcting me.
I just wish there was an easy way to define a list as list<?> and to run
queries against. But, sounds like there isn't (and nobody is seeing need
for it) so I think I'll just take one of the suggested workarounds...
Tuukka
On Wed, Jun 25, 2014 at 10:47 AM, Sylvain Lebresne <sylv...@datastax.com>
wrote:
> On Wed, Jun 25, 2014 at 8:49 AM, Tuukka Mustonen <
> tuukka.musto...@gmail.com> wrote:
>
>> Unfortunately, I need to query per list items. That's why I'm running
>> Cassandra 2.1rc1 (offers secondary indexes for collections).
>>
>
> Using a list of blobs does not in any way prevent you from doing that.
> Types are constraints on what values C* will accept and using blob is
> simply asking C* to not reject any value. Doing so does not in any way
> limit the kind of queries you can do.
>
> The small downside of using blobs is that you'll have to
> serialize/deserialize your value manually client-side, but that's not a
> huge deal either. That said, if you really only have 3 types of values to
> store and if you don't particularly care about the order of items in the
> collection (i.e. if you said you want a list but could really do with a
> set), then storing 3 different sets can be a viable solution too (as in,
> there is no strong downside to doing it as far as C* is concerned and it
> may be simpler to deal with client side (or not, it depends a bit on what
> your client side code does exactly)).
>
>
>>
>> As I understood it, also Cassandra supports dynamic schemas, but only
>> through Thrift protocol.
>>
>
> "dynamic schemas" is a terribly imprecise term that means different things
> to different people, but in general that statement is incorrect: you can do
> the same things with CQL and with Thrift.
>
>
>> Also, I don't think it changes the fact that collections need to be
>> strongly-typed in Cassandra, no matter what protocol is used?
>>
>
> Well, yes since you do have to provide a type for the elements in the
> collection, but as said previously that does not in any way prevent you for
> having "collections of anything" since you can use a blob type.
>
> --
> Sylvain
>
>
>>
>> Tuukka
>>
>>
>>
>> On Tue, Jun 24, 2014 at 9:41 PM, DuyHai Doan <doanduy...@gmail.com>
>> wrote:
>>
>>> "Jeremy, with blob field (ByteBuffer), I can query exact matches (just
>>> encode the value in query), but greater/less than queries would not work.
>>> Any sort of serialization kills "native" ways to query data" --> Not
>>> necessarily. You still use "normal" types (uuid, string, timestamp,...) for
>>> clustering columns and use them for querying. For the cells where you store
>>> values, use blob type.
>>>
>>>
>>>
>>>
>>> On Tue, Jun 24, 2014 at 8:21 PM, Tuukka Mustonen <
>>> tuukka.musto...@gmail.com> wrote:
>>>
>>>> What if I need to query by list items?
>>>>
>>>> 1. Jeremy, with blob field (ByteBuffer), I can query exact matches
>>>> (just encode the value in query), but greater/less than queries would not
>>>> work. Any sort of serialization kills "native" ways to query data
>>>> 2. Even with user defined types, I would need to define separate fields
>>>> for each value. Running queries would be cumbersome (something like WHERE
>>>> items CONTAINS {'text_value': 'foobar'} or WHERE items CONTAINS
>>>> {'int_value': 3}. Pavel, did you mean like this?
>>>>
>>>> I'm running 2.1rc1 with python driver 2.0.2.
>>>>
>>>> Tuukka
>>>>
>>>>
>>>> On Tue, Jun 24, 2014 at 4:39 PM, Pavel Kogan <pavel.ko...@cortica.com>
>>>> wrote:
>>>>
>>>>> 1) You can use list of strings which are serialized JSONs, or use
>>>>> ByteBuffer with your own serialization as Jeremy suggested.
>>>>> 2) Use Cassandra 2.1 (not officially released yet) were there is new
>>>>> feature of user defined types.
>>>>>
>>>>> Pavel
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On Tue, Jun 24, 2014 at 9:18 AM, Jeremy Jongsma <jer...@barchart.com>
>>>>> wrote:
>>>>>
>>>>>> Use a ByteBuffer value type with your own serialization (we use
>>>>>> protobuf for complex value structures)
>>>>>> On Jun 24, 2014 5:30 AM, "Tuukka Mustonen" <
>>>>>> tuukka.musto...@gmail.com> wrote:
>>>>>>
>>>>>>> Hello,
>>>>>>>
>>>>>>> I need to store a list of mixed types in Cassandra. The list may
>>>>>>> contain numbers, strings and booleans. So I would need something like
>>>>>>> list<?>.
>>>>>>>
>>>>>>> Is this possible in Cassandra and if not, what workaround would you
>>>>>>> suggest for storing a list of mixed type items? I sketched a few (using
>>>>>>> a
>>>>>>> list per type, using list of user types in Cassandra 2.1, etc.), but I
>>>>>>> get
>>>>>>> a bad feeling about each.
>>>>>>>
>>>>>>> Couldn't find an "exact" answer to this through searches...
>>>>>>> Regards,
>>>>>>> Tuukka
>>>>>>>
>>>>>>> P.S. I first asked this at SO before realizing the traffic there is
>>>>>>> very low:
>>>>>>> http://stackoverflow.com/questions/24380158/storing-a-list-of-mixed-types-in-cassandra
>>>>>>>
>>>>>>>
>>>>>
>>>>
>>>
>>
>
