Log

• Get rid of the x/w aliases to value/count by Caio 8 years ago
• Get rid of TDigest._{count,mean} aliases 💬 by Caio 8 years ago

  This patch gets rid of the _count() and _mean() helpers and move
  them into the summary package. What I really wanted were macros,
  but hey ¯\_(ツ)_/¯
  
  Summary is still leaky, but at least TDigest only uses its
  public interfaces now.
  

• Rename `{}summary.keys` to `means` 💬 by Caio 8 years ago

  $ gorename -from '"github.com/caio/go-tdigest".summary.keys' -to means
  

• Completely rework the quantile estimation codepath 💬 by Caio 8 years ago

  This patch is too big, but there isn't much getting away from it
  in smaller steps because summary{} and TDigest{} are actually
  tightly coupled (i.e.: the abstraction is mostly useful for
  code organization but fails at isolation).
  
  The major changes are:
  
  - Summaries now hold repeated items instead of just unique means
    and their respective counts (which led to changes in how the
    digest adds new centroids too)
  - Quantile estimation is now a straight port from the reference
    implementation (issue-84 branch)
  
  The digest now doesn't potentially report completely wrong values
  on distributions with multiple steep hills nor biases in favour
  of big centroids with few occurrences.
  
  This patch closes #12. Some historical details for the motivation
  for this work can be found on PR #11.
  

• Make TestMerge more thorough 💬 by Caio 8 years ago

  This patch makes the test use multiple partitioning configurations and
  a lot more items (so that we can partition and merge more). This in
  turn means that the whole test is significantly slower (still very
  tolerable though) but helps uncover subtle drifts where precision
  is lower.
  

• Output more details when TestRespectBounds fails by Caio 8 years ago
• Add TestSingletonInACrowd 💬 by Caio 8 years ago

  Notice that this test has its exterme quantile (0.999)
  skipped because the java reference implementation behaves
  the same. Reasoning and more details on issue #12
  

• Drop TestNonUniformDistribution 💬 by Caio 8 years ago

  I know it's weird to drop failing tests, but bear with me:
  
  This test was added on 011e706e and has worked nicely since
  then, however I'm having a hard time accepting the error
  thresholds for this manually crafted distribution. Given
  that it fails with the java implementation as well (AVL
  and Merging -based versions), I'm considering it bugged.
  
  I'll revisit this one day, but it looks like to me it
  would be more productive to actually test distributions
  and their relative errors in a standard manner as in
  TDigestTest.java
  
  --- FAIL: TestNonUniformDistribution (0.00s)
          tdigest_test.go:85: T-Digest.Quantile(0.2500) = 420.1363 vs actual 499.9531. Diff (79.8168) >= 11.0000
  

• Make travis CI test with go 1.9, drop 1.6 by Caio 8 years ago
• Merge pull request #11 from christineyen/master 💬 by Caio 8 years ago

  Ensure returned values stay within bounds

• Ensure returned values stay within bounds by Christine Yen 8 years ago
• Update README with project status by Caio 9 years ago
• Test against Go 1.8, stop testing vs tip 💬 by Caio 9 years ago

  Go 1.8 has been released. Testing tip is too slow for my taste.
  

• Revert "gometalinter: Use interfaces when it makes sense" 💬 by Caio 9 years ago

  This reverts commit 2686af5cd15f0a8f4650e2a4a9bbf44c05b87e10.
  
  I still don't like this even after sleeping on it :-)
  

• Reference `gopkg.in` in the README by Caio 9 years ago
• Add syntax highlighting to the README example 💬 by Caio 9 years ago

  Look at this with the `-w` switch and you'll see that all that
  has changed is that I wrapped the example code with some "```"
  and adjusted the indent.
  

• gometalinter: Use interfaces when it makes sense 💬 by Caio 9 years ago

  There's now another suggestion:
  
      $ gometalinter ./... --disable=gocyclo
      serialization.go:63:16:warning: buf can be jpeg.Reader (interfacer)
  
  But that's just too silly.
  
  To be honest, I'm not so sure this is a sensical change even:
  encodeUint and encodeUint are "private". bytes.{Buffer,Reader} is
  precisely what I need and use throughout the implementation. So,
  hashtag revertMeMaybe.
  

• gometalinter: Error checking on tests 💬 by Caio 9 years ago

  With this patch we now explicitly ignore most errors coming from Add()
  (as there are specialized tests for it and the consistency checks
  cover the error scenario) and introduces a check for potential Merge()
  and Compress() errors.
  
  Not really happy with the mass ignore, but I do like the `errcheck`
  linter, so there's that.
  

• gometalinter: Simplify if-return clause by Caio 9 years ago
• gometalinter: Avoid redefining `c` 💬 by Caio 9 years ago

  Harmless, but an easy fix anyway :)
  

• Add missing error checks 💬 by Caio 9 years ago

  This patch fixes some missing error checks as pointed out by
  `gometalinter` (thanks @dgryski!).
  
  There is a signature change on Merge() and Compress(), from returning
  nothing to now returning an error type.
  

• Remove the gocover.io banner 💬 by Caio 9 years ago

  Too damn unstable for my taste.
  

• Test explicitly against 1.7, drop 1.4 by Caio 9 years ago
• No need for a tmp var to make a swap by Caio 10 years ago
• No more `t.Parallel()` anywhere by Caio 10 years ago
• Stabilize the rng to the Sequential Insertion test 💬 by Caio 10 years ago

  Before this patch I would see the occasional failure when running:
  
      $ go test -run Sequential -count 30
  
  The failures are consistently _not_ on the extreme quantiles, as
  expected given the data-structure characteristics. I guess this is
  because `Add()` and `shuffle()` (which is used by `Compress()`, that
  gets triggered several times for this test given the small summary
  size) use the same rng source, but I'm not even remotely close to
  someone who properly understands all this math.
  
  Shouldn't be a real problem, but I'll consider adding support for
  providing a random source instead of using the default one so that
  at least we get no contention from multiple things asking for a random
  number to the global source.
  

• Revert "Try out this travis-ci + coveralls thing" 💬 by Caio 10 years ago

  This reverts commit 42ef656694f47a21c37295c1ae3446666ee04c1f.
  
  LOL This randomly fails with:
  
      Bad response status from coveralls: 422 - {"message":"Couldn't find
      a repository matching this job.","error":true}
  
  And here shows that there's something real silly going on:
  
      https://github.com/lemurheavy/coveralls-public/issues/487
  
  So no, not worth the pain. $fancy--
  

• Try out this travis-ci + coveralls thing by Caio 10 years ago
• Set count to zero after creating a new summary by Caio 10 years ago
• Assert we don't change counts during Compress() 💬 by Caio 10 years ago

  So that we avoid causing a regression (Ref: PR #10)