Request: unique hash on query objects for logging

I’m trying to write a simple metrics tool to measure which queries are performing slowly so that I can evaluate my codebase over time. As a simplified example:

var knex = require('knex')({myConfig});

var runningQueries = {};

knex.on('query', function (data) {
  // Generates a unique has using hash using keys like:
  // ['__knexUid', 'method', 'sql', 'bindings', 'options']
  var queryHash = generateHash(data);
  runningQueries[queryHash] = {
    sql: data.sql,
    start: Date.now()
  };
});

knex.on('query-response', function (response, obj, builder) {
  // This hash should match the one in the function above.
  var queryHash = generateHash(obj);
  var runTime = Date.now() - runningQueries[queryHash].start;
  // etc. etc.
  delete runningQueries[queryHash];
});

The problem I’m encountering is that multiple queries with the same properties will get jumbled together - and could potentially screw up my timings or cause me to miss queries. It would be ideal if each query object had a unique hash attached to it (or the Date.now() timestamp it was created at) to identify them uniquely.

I can add the code, but before I submit a PR I’d like your input and to make sure I’m not wasting your time. 😃

[funnylookinhat]

@wubzz That’s a good idea. I’ve gone ahead and updated it to __knexQueryUid to fit the __knexUid pattern.

[E-Kuerschner]

I’m using knex@0.21.2 with mssql

I’ve noticed that some queries don’t receive a __knexQueryUid. Any idea why that might be? It appears that anything using the schema api is missing this, for example:

await connection.schema.dropTableIfExists(tableName)

this type of code is resulting in query event data looking like:

{
  __knexUid: '__knexUid3',
  __knexTxId: undefined,
  sql: "if object_id('***', 'U') is not null DROP TABLE ***",
  bindings: []
}