Can query a node from index

[hansman]

What happens here?

• create a node
• save the node
• index the node
• get the indexed node

var indexName = 'myNodeIndex';
var node = db.createNode({name: 'hey'});
node.save(function() {
db.index(indexName, 'name', 'hey', function() {
db.getIndexedNode(indexName, 'name', 'hey', console.log)
})
})

Expected:
getIndexedNode returns the node that just got indexed.

Actual:
getIndexedNode has empty response

[aseemk]

You're passing console.log as a callback directly. Can you instead pass a proper function (err, node) { console.log(node); } callback? Does that work?

[aseemk]

The other thing is, you're not seeing if you're getting any errors. For each of the callbacks, it's good to check if (err) throw err; where err is the first param.

[hansman]

For sure. This code is only to point out the flow. console.log is sufficient to see that there is an empty response which is the point of this demonstration.

[aseemk]

Got it. Would you mind confirming a couple things for sanity:

• You write db.index, but your code is actually node.index, right?
• There are no errors returned for any of these calls, incl. node.index?

Assuming both of these are indeed the case:

• What happens if you check the index after some delay (e.g. a few seconds)? Does it show up? If so, it would seem that Neo4j legacy indexes are eventually consistent, not immediate. I don't know anything about that, though; might check with the Neo4j mailing list or Slack.
• You might already be aware of this, but could you use Neo4j v2 schema indexes? Those are immediately consistent, I'm pretty sure.

[hansman]

no errors. yes I use node.index. I can't get the node from the index even after a long delay. Not a eventual consistency case.

[hansman]

question: have you tried doing this yourself?

[aseemk]

Okay, surprising.

No, I haven't tried this myself. We use exclusively v2 schema indexes and legacy auto-indexes, not legacy manual indexes. I can try myself later.

Another thing to try would be using the Neo4j webadmin to do these things manually. Does it work there? If so, suggests a node-neo4j or code bug. If not, suggests a Neo4j bug.

[aseemk]

However, we do cover manual indexing in our tests, and that's always worked:

node-neo4j/test/crud._coffee

Lines 220 to 224 in b1f6348

	'index nodes': (_) ->
	daniel.index 'users', 'name', 'Daniel', _
	node = db.getIndexedNode 'users', 'name', 'Daniel', _
	expect(node).to.be.an 'object'
	expect(node.exists).to.be.true