Remove diffSample optimisation

[adinapoli-iohk]

@23Skidoo @tibbe This PR removes the diffSample optimisation to allow statsd to pick us these "low-level" details for us, as client code shouldn't be bothered.

See here for a better description and the rationale around this PR: #16

[adinapoli-iohk]

The Travis error seems to be a red herring:

Errors were encountered while processing:
 /var/cache/apt/archives/cabal-install-head_2.1+git20170813.0.267efc8-4~14.04_amd64.deb
 /var/cache/apt/archives/ghc-8.2.1_8.2.1-2~14.04_amd64.deb
W: --force-yes is deprecated, use one of the options starting with --allow instead.
E: Sub-process /usr/bin/dpkg returned an error code (1)

[tibbe]

Probably goes without saying, but please do test these changes against statsd to make sure you get the expected behavior.

[adinapoli-iohk]

Sounds like a plan. I'm not expert enough with statsd to judge whether or not this PR is solving our woes, but let me get back to you once we have done an internal run with the feature branch this PR was created from. cc @domenkozar

[23Skidoo]

LGTM. Can you please add a note to the changelog?

[domenkozar]

@adinapoli-iohk let's do an upstream stack change to use your branch and we'll deploy that to staging to verify.

[adinapoli-iohk]

@domenkozar Absolutely.

LGTM. Can you please add a note to the changelog?

Sure thing, leave it with me. I will confirm this is working as intended with Domen and will update the changelog accordingly. Which version should I put into the changelog? CURRENT_VERSION + 1 I guess ?

[23Skidoo]

0.2.2.0, I think: even though this change is backwards-compatible, it changes the behaviour, so it should be possible to guard against it with a MIN_VERSION_ macro.

[domenkozar]

I can confirm we see multiple consequent same-value pushes to statsd using this PR.

[adinapoli-iohk]

@domenkozar Neat! I guess now it's my turn to update the changelog so that we can make the PR officially mergeable 😛

[23Skidoo]

Merged, thanks!

[23Skidoo]

ekg-statsd-0.2.2.0 is now on Hackage.