Concerns about the way node-oracle handles cursors in results from stored procs

[richsad]

While reviewing the C++ code for my previous issue #61, I noticed that when a stored proc returns a cursor the node-oracle code in connection.cpp automatically iterates over the results and builds an array of all results. This is fine for smaller datasets and even datasets with thousands or tens of thousands of results, but it would be a problem for any stored proc returning millions of rows.

My guess is that the cursors are hidden from node.js for simplicity reasons. But this eventually going to blow up in someone's face. It would be awesome to have a second method for calling a stored proc that returns a cursor to pass some object representing the cursor back to calling node.js code and allow the node.js programmer to then iterate over the results.

While I will most like use the fork I made today to fix my problems in issue 61, I doubt I will have time to tackle the results cursor. Has anyone ever encountered a problem with the returned array? As I've mentioned I am doing work for a very large media company. There are certainly values that we may query from node.js that could approach a million rows!

[brandonculver]

This would be ideal, would like to interact with a table that has 4.5 million rows.

[bjouhier]

I need this too. Today I'm handling it with a window which is handled in JS but it only works if the select is sorted on a unique index.

I implemented a similar thing in node-db/node-db-oracle (mariano/node-db#3). I was going to do the same thing with node-oracle unless someone already has it in store.

[bjouhier]

We can probably close this one now.