Remove previous and next cache in SearchResult

[dbengsch]

Fix #176

• Remove _previous and _next cache in SearchResult
• Remove previous function that could't be reworked properly for a scroll

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (develop@138d393). Click here to learn what that means.

@@            Coverage Diff             @@
##             develop     #177   +/-   ##
==========================================
  Coverage           ?   99.33%           
==========================================
  Files              ?       16           
  Lines              ?     1642           
  Branches           ?      439           
==========================================
  Hits               ?     1631           
  Misses             ?       11           
  Partials           ?        0

Impacted Files	Coverage Δ
src/Collection.js	97.37% <100%> (ø)
src/SearchResult.js	97.05% <94.73%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 138d393...c3f7e87. Read the comment docs.

[ballinette]

???

[dbengsch]

This states the object built by the constructor function has a defined property called dataCollection of type Collection

[ballinette]

OK, my bad. I read "@param" instead of "@Property"

[xbill82]

PR needs a more detailed description. I found no mention about next and previous search results in the docs.
Also don't understand why we had these before and why we take them out now.

[dbengsch]

@xbill82:
Issue about the documentation created kuzzleio/documentation#88 .

[dbengsch]

@xbill82 Please refer to the connected ticket to know why we had to remove them.

[xbill82]

@dbengsch Thanks, I actually didn't really read the title of the issue (just got annoyed by the "no description" message, but the title is actually enough :) I understand that the problem is that they introduce a memory leak, but I've missed all the discussion that lead to the decision to take them out instead of fixing the mem leak. We can discuss it IRL if u prefer.

[benoitvidis]

@xbill82 This issue is severe enough imo to take immediate actions and remove this pseudo-functionality just right now.

Users can very easily work around the removed previous options on client side by storing the whole dataset in a local collection but this is not what Kuzzle should do on its side by default.

[scottinet]

@xbill82 > this is not a memleak, since the used memory is not lost, but memory consumption. And that's the problem: there is no way to implement this feature without the client consuming too much memory.

It has been decided to make this cursor "forward-only", as are many cursor implementations out there.

[xbill82]

@scottinet thanks for the details, now I get the point (and the severity) of the issue.