>> I want to do something that is perfectly satisfied by an hstore column. *Except* that I want to be able to do fast
(ieindexed) <, > etc comparisons, not just equality.
>>
>> From what I can tell, there isn’t really any way to get hstore to do this, so I’ll have to go to a key-value table.
ButI thought I’d ask just in case I’m missing something.
>>
>
> I think your missing something.
>
> Is it one field in the hstore? Did you try an expression index?
>
> create index cars_mph on cars ( (data->'mph') );
>
> thats a btree index, which should support < and >. (Although I've never tried it)
Sorry I wasn’t clear. I need a fast comparison on whatever keys someone chooses to put in the hstore.
I’m creating a web service where you can post whatever keys and values you wish. I am leaning toward having folks
declarethe relations they want to store and the field types, so I could just create a table for each one, but I want
thisto be able to scale to a lot of data for a lot of people.
Perhaps I could give everyone their own schema and just create tables for their relations. How heavyweight are schemas?
But if I don’t want to do that, I’m leaning toward a table with a relation name and a hstore, and then separate k-v
tablesfor different types. I was thinking of keeping the hstore because they will often be searching on fewer fields
thanthey’re retrieving, so this would avoid having to do a join for every field they need.
Regards,
Guyren G Howe
Relevant Logic LLC