incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Wout Mertens <wout.mert...@gmail.com>
Subject Re: group_level and sorting
Date Mon, 20 Apr 2009 21:04:07 GMT
On Apr 20, 2009, at 4:30 PM, Nicolas Clairon wrote:

> On Mon, Apr 20, 2009 at 3:39 PM, Wout Mertens  
> <wout.mertens@gmail.com> wrote:
>> On Apr 20, 2009, at 9:51 AM, Nicolas Clairon wrote:
>> Hmmm... Remember that reduce results should not grow faster than  
>> log(n) for
>> n documents. If everyone uses a different tag, that can become a  
>> problem. Of
>> course, if you're not expecting thousands of tags, that shouldn't  
>> be a real
>> issue.
>
> I don't expect thousands of tags but it might be possible that each
> document is so
> different that I can have "nb_results_map" = "nb_results_reduce" (if,
> for instance,
> the description is different for each document).
>
> How can I handle this issue ?

Err, no that's not an issue. The issue is when a single result row  
grows proportionally to the number of documents that are processed in  
that result.

Tags are probably ok.

You can also map: emit ([doc.title,doc.tag],1) and reduce:sum(v) to  
generate a list tags that a certain document uses and how popular each  
tag is. You'd have to query that list separately.

Wout.

Mime
View raw message