CBL 2.0 Query & Join Syntax Question

Mobile Couchbase Lite
1

Hi,

we are working with CBL 2.0 .NET and seemingly I can’t figure out how to correctly use Joins in Queries.

Consider a database with two simple documents:

{
	"_id": "author_1",
	"name": "John Doe"
}

–

{
	"_id": "blog_post_1",
	"title": "Lorem Ipsum",
    "author": "author_1"
}

Since the documentation for Joins is not up yet, I tried to fiddle around and thought the following code should work (db being a reference to the CBL-Database):

var query = Query.Select(SelectResult.All())
			.From(DataSource.Database(db))
			.Joins(
			    Join.InnerJoin(DataSource.Database(db)).On(
                    Expression.Meta().ID.EqualTo(Expression.Property("author"))
                )
			);

IResultSet rows = query.Run();

But running that Query never finishes. It also throws no exception or such, so I don’t really know what I’m doing wrong. I also tried a couple of other things, like switching the “JoinOn”-Expression the other way around as well as joining on properties instead of meta-data. But the query always just never finishes.

So I hope someone could give me a pointer on what the correct API-Usage would be :).

As a side-question I would also ask if this is “the way to go”. NoSQL is still kind of new to us.
In reality we are talking about ~1-5 million documents of various “types” (i.e.: each “type” being akin to a SQL-Table and each document representing something akin to a SQL-Row-Entry and we put all documents from all tables into 1 bucket). Would you use N1QL-Joins like above in such a scenario or would it be faster/better to use GetDocument(authorId) instead? Or even use views instead?

Thanks & best regards,
Lars

2

But running that Query never finishes.

How many documents in this database? Are the CPU or disk heavily loaded? If not, can you pause the app and get a backtrace?

Your Join clause doesn’t specify which sources the ID and “author” properties belong to. I’m not actually sure what happens in that case; it depends on how SQLite handles the resulting translated query. You should use the “As” operator to name both the From source and the Join, and then use those names. (Sorry, I don’t know the exact CBL syntax for this. @pasin, can you help?)

But of course, even with this problem your query should still finish, with or without an error, so this looks like a bug…

3

The bucket currently contains 21 documents. So I’d expect a more or less instantaneous response ;).

Yeah, I was wondering on how to assign id and author to the respective “document-sets”, but couldn’t figure it out. If this is the problem and someone could help me out with the correct API-Usage, that would be super nice :).

It seems like a bug, as in “it should throw an error”. But since 2.0 is still in development I’d assume that wrong API-usage might not yet be handled correctly in some cases :).

[Edit]: When I’m back in the office and if this is a bug, I’ll post a bug-report. Last time I checked, symbol-server weren’t working though, so maybe no backtrace (but I’ll try again).

4

Hmmm I think it should be throwing an error without having the required AS clauses in the join. In lieu of proper documentation I suggest looking at the unit tests for some guidance. There is a test called TestJoin which should be useful.