startkey_docid and endkey_docid are either undefined, or strings containing a doc id. As such, they shouldn't be parsed. This is already the behavior of the real implementation. The mock implementation is failing my tests, although it works with the real one.

The change for group_level is only a simplification in JS.

The way keys are compared in the Bucket Mock results in a wrong sort order for compound keys with integers.

In Mock mode I get view results like that:

[1,1]
[1,10]
[1,100]
[1,2]
[1,3]

Currently the only way to operate on document flags is to implement the bucket-wide transcoder functions. However when the application itself already handles encoding/decoding (in our case for compatibility between many data stores), the transcoder functions become more of a nuisance than a feature. i.e. we already have serialised UTF-8 or Binary data and know what the data type should be, e.g. JSON, in this case we would like to set the flags directly as was available in couchnode v1.x.

This pull request re-implements this feature by allowing you to pass in the flags options arguments for all store operations and returns it as the last argument of get operation callbacks.

I tried to make the implementation as minimal and backwards compatible as possible.

Yo,

this adds io.js 1.0.X support and bumps to Nan 1.5.1 (adds 0.11.XX support in terms of node-legacy).

_RFC._

jfyi: The "Orphan:" prefix indicates that there is neither a ticket for this change nor do I want to create an Issue addressing this change.

According to https://docs.couchbase.com/server/6.0/fts/fts-geospatial-queries.html#specifying-distances Radius-Based is like following

{
  "from": 0,
  "size": 10,
  "query": {
    "location": {
      "lon": -2.235143,
      "lat": 53.482358
     },
      "distance": "100mi",
      "field": "geo"
    },
  "sort": [
    {
      "by": "geo_distance",
      "field": "geo",
      "unit": "mi",
      "location": {
      "lon": -2.235143,
      "lat": 53.482358
      }
    }
  ]
}

ESlint is a newer pluggable linter that does both the job of JSHint and JSCS (the JSCS team has actually joined the ESLint team)

This PR is both a translation of the existing JSHint rules and a set of fixes based on issues found. It is structured so that if there is a rule you want changed it should be easy for me to rollback the commit and update the .eslintrc

Let me know if you have any questions or want any rule changes

This enables iojs 2.x.x it is just a very small change to remove a warning and a nan bump. Honestly would probably be easier to just just redo yourself but needed it for my work so figured I would submit

As explained in this issue every multi method has been removed from couchnode (only getMulti still exists). This commit is a proposal to reintroduce them — but only as helpers, this implementation only wraps existing code and do not use libcouchbase directly.

Once in a while, using an upsertMulti can be useful, so instead of letting everyone re-implement it all over again, it seems appropriate to offer it with couchnode.

This implementation relies heavily on the previous getMulti implementation. A _multi method gather every bit of code while each operation (upsert, insert, touch) only calls it with the proper arguments. This _multi method handles validation, options (single's options override multi options) and produces a standard output.

The code tries to stick as much as possible to the existing coding standards and a test + a documentation is written for every bit of it.

As an example:

bucket.getAndLockMulti([key1, key2, key3], function(err, v) {
  assert(err === 1);
  assert(v[key1].cas);
  assert(v[key1].value);
  assert(!v[key1].cas);
  assert(v[key1].error);
})

var kv = {};
kv[key1] = {value: 'foo'};
kv[key2] = {value: 'bar', expiry: 5};
bucket.upsertMulti(kv, {expiry: 10}, function(err, v) {
  assert(err === 2);
})

ps. was not able to submit this via review.couchbase.com, has sign the Contributor Agreement, but got no mail

ESlint is a newer pluggable linter that does both the job of JSHint and JSCS (the JSCS team has actually joined the ESLint team)

This PR is both a translation of the existing JSHint rules and a set of fixes based on issues found. It is structured so that if there is a rule you want changed it should be easy for me to rollback the commit and update the .eslintrc

Let me know if you have any questions or want any rule changes

Hi,

My couchbase-backed hapi server kept crashing with "Got memcached EINVAL!" as described on the couchbase forums.

According to a link in the forum thread above the issue was coming from libcouchbase and has been fixed in version 2.5.8.

So I've updated the dependency in couchnode to libcouchbase 2.5.8, and it solved my issue.

To be honest, I don't perfectly understand what I did, being quite unfamiliar with node-gyp :)

That being said, I've been able to run the test suite successfully both before and after the update of the dependency (using the real bucket, too), so I guess this contribution is at least a good base for discussion.

Let me know what you think!

You shouldn't throw errors inside functions that use callbacks. This pull request isn't extensively tested, but the existing tests pass and it's certainly a step in the right direction.

Bumps json5 from 2.1.3 to 2.2.3.

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.