Using indexed properties bundles

[Jason Dentler]

Hi,

I couldn't find any documentation, but I think this bundle does what I need.

I have 3 types of documents: communities, floor plans, and model homes. FloorPlans and ModelHomes both have a Price and a CommunityId property. I need to display "Homes starting at $222,900" on the community's page and in a lot of community search results.

I have an index to get the minimum price per community. I think this bundle will automatically keep my community doc's MinimumPrice property accurate. How do I configure the bundle to make that magic happen?

- J

[Oren Eini (Ayende Rahien)]

The best "docs" are here:

http://issues.hibernatingrhinos.com/issue/RavenDB-245 

We are working on that.

Note that for what you need, I don't see why you would need this.

Can you explain a bit more why you want to put it in the community doc?

What sort of searches are we talking about?

Let us assume that you have the following index:

communities/MinPrice

from h in docs.Homes

select new { h.CommunityId, MinPrice = h.Price } 

from result in results

group result by result.CommunityId into g

select new { CommunityId = g.Key, MinPrice = g.Min(x=>x.MinPrice) }

You would then create the following document:

Raven/IndexedProperties/communities/MinPrice 

{

  "DocumentKey" : "CommunityId",

  "FieldNameMappings": {

          "MinPrice" : "CommunityMinimumPrice"

[Jason Dentler]

Thanks Oren! That's exactly what I need. The indexed property bundle lets me get this base price calculation out of the way of our already-complex search requirements.

There are simple community lists all over the site, but those aren't a concern. We have one "Find a home" page where the criteria can really get out of hand. An example might be "Show me homes in Northwest Houston under $310,000 with at least 4 bedrooms and a 2 car garage, with a golf course and pool nearby"

The city, part of town, and nearby amenities are stored in the community doc and the rest is in the floor plan docs.

The displayed results would be something like this:

Bridgeland, from the $240s, 2500-5000 sq ft.

 - The Beezer, from the $260s, 2600 sq ft, 4 bedrooms, 2.5 baths, 2 car garage

 - The Barker, from the $300s, 3300 sq ft, 4 bedrooms, 3 baths, 2 car garage

The Cove, from the $190s, 2400-3400 sq ft

 - The Bellville, from the $200s, 2600 sq ft, 4 bedrooms, 2 baths, 2 car garage

I expect the actual results from Raven will be some flat composite of community & floor plan data sorted by community. We'll do the grouping as we loop through the result.

We're not sure if this one complex search page is something we should even be doing with Raven. We're also considering doing the more complex filtering in memory, or replicating to SQL server and querying there. We haven't tried any of it yet, but if you've got ideas...

- J

[Oren Eini (Ayende Rahien)]

What is the query you are issuing?

You shown the results, but not the actual query

[Jason Dentler]

Oren,

We haven't built that query yet. We're still prototyping & evaluating if RavenDB is the right choice for this project. 

An typical search might be "Show me homes in Northwest Houston under $310,000 with at least 4 bedrooms and a 2 car garage, with a golf course and pool nearby"

- J

[Oren Eini (Ayende Rahien)]

Show me homes in Northwest Houston under $310,000 with at least 4 bedrooms and a 2 car garage, with a golf course and pool nearby 

The yellow things are data that belongs to the house, right?

And the green to the community?

[Jason Dentler]

You got it.

By the way, the indexed properties bundle disappeared from your 1.2 branch. Why?

- J

[Jason Dentler]

Just to clarify, the user is required to search within a market (Houston, in this example). The other search criteria are optional.

- J

[Kijana Woodard]

Hmmmm. I wonder if Facets may be of value to you.

[Oren Eini (Ayende Rahien)]

Yes, that was what I was thinking.

You probably want to start with a search to get the market.

Then, you search for homes in that market, and provide the user with facets for them.