That won't work for a compound key - Pulkit wants to sort ascending on the 
first item in his key and descending on the second. Paul is talking about new 
functionality that augments the descending=true option. 


On Wednesday, 5 September 2012 at 07:29, Andrey Kuprianov wrote:

> It's not per view, but per single query. Take a look here for more
> details:
> http://wiki.apache.org/couchdb/HTTP_view_API#View_Generation_Options
> 
> On Wed, Sep 5, 2012 at 1:42 PM, Paul Davis <[email protected] 
> (mailto:[email protected])>wrote:
> 
> > Unfortunately no. There was a patch to define a sort order for array
> > keys but it was never committed to master. Its not a terribly
> > difficult patch but there hasn't ever been much of a call for such a
> > thing (AFAIK you're the second ever) and it has some awkward edge
> > cases. Perhaps a revisit would be kosher.
> > 
> > For background the original patch was a design doc member of [true,
> > false, true, true] (["asc", "desc", "asc", "asc"] would be
> > equivalently easy). We'll need to look into making this a per-view
> > option as well as including it in the design doc signature. We'd also
> > need to account for the newer raw collation option as well and define
> > that. Or perhaps make this a new collation option.
> > 
> > If anyone wants to dig into this I can point to the major code areas
> > and general points of concern but I don't have time to dig in
> > immediately.
> > 
> > On Tue, Sep 4, 2012 at 11:16 PM, Pulkit Singhal <[email protected] 
> > (mailto:[email protected])>
> > wrote:
> > > I have a map with a compound key which naturally sorts in an ascending
> > > order as:
> > > [No, 2011]
> > > [No, 2012]
> > > [Yes, 2011]
> > > [Yes, 2012]
> > > 
> > > But I want it to sort as:
> > > [No, 2012]
> > > [No, 2011]
> > > [Yes, 2012]
> > > [Yes, 2011]
> > > 
> > > Which means the first element in the key No/Yes should be sorted in an
> > > ascending order, whereas the timestamp should be sorted in a descending
> > > order.
> > > 
> > > Is there a way to do this in a query? Or do I have to come up with some
> > > hack in my mapping function?
> > > 
> > 
> > 
> 
> 
> 


Reply via email to