treojs / treo-websql Goto Github PK

Using treo-websql, we've noticed the warnings about index being inefficient, and looking at it see that it uses a cursor. However, in shim.js, find converts an IDBKeyRange to a (presumably somewhat more efficient) WHERE clause.

What are the reasons for using cursor to implement index over using a WHERE clause? Is it just a matter of writing code, or are there other considerations of which I am ignorant?

While trying to figure out why our queries were failing in Safari I noticed that this function does not cater for multi-field indexes when converting an IDBKeyRange for use with WebSQL.

function testValue(v) {
  return (((!r.lowerOpen && v >= r.lower) || (r.lowerOpen && v > r.lower)) && ((!r.upperOpen && v <= r.upper) || (r.upperOpen && v < r.upper))
    || (r.upper === undefined && ((!r.lowerOpen && v >= r.lower) || (r.lowerOpen && v > r.lower)))
    || (r.lower === undefined && ((!r.upperOpen && v <= r.upper) || (r.upperOpen && v < r.upper))));
}`

I suspect that using this function to translate IDBKeyRanges as mentioned in #1 will actually fix this.

If #1 is not a simple fix or won't work for these issues I'm happy to submit a PR to allow testValue to support multi-field IDBKeyRanges.