Commits

Author Commit Message Labels Comments Date
Bryan O'Sullivan
Sort extra-source-files and data-files
Bryan O'Sullivan
Merge pull request #39 from nomeata/master Fix for #37
nomeata
Actually ship the JS source files in sdist
nomeata
Add sources for javascript files and replace templates/js/excanvas-r3.min.js by the file found in excanvas_r3.zip, to ensure that licence conditions are fulfilled. This closes #37.
Bryan O'Sullivan
Merge
Bryan O'Sullivan
Added tag 0.8.0.1 for changeset b983f58b067c
Bryan O'Sullivan
Bump version to 0.8.0.1
Tags
0.8.0.1
Bryan O'Sullivan
Add upper bound on statistics
Bryan O'Sullivan
Tiny style tweaks
Bryan O'Sullivan
Merge pull request #38 from snoyberg/master binary 0.5 support
Michael Snoyman
Added some missing Binary instances #38
Michael Snoyman
Merge branch 'criterion-0.8'
Michael Snoyman
binary 0.5 support
Bryan O'Sullivan
Bare minimal skeleton app
Bryan O'Sullivan
Use new Estimator type
Bryan O'Sullivan
Merge
Bryan O'Sullivan
Bump dependency on statistics
Gabriel Gonzalez
Fixed to build against `cassava-0.3.0.0`. Fixes #34 `cassava-0.3.0.0` changed `encode` to accept a list instead of a vector. This patch changes `Criterion.IO.Printf` to use the new `encode`, and also sets a lower bound of `cassava-0.3.0.0`.
Bryan O'Sullivan
Ignore cabal sandbox stuff
Bryan O'Sullivan
Bump version to 0.10.0.0
Bryan O'Sullivan
Fix normal form doco (gh-31)
Bryan O'Sullivan
Produce valid CSV (gh-23) The previous method of dumping out CSV data was a grotesque hack. This is waaaaay more sane.
Bryan O'Sullivan
Oops
Bryan O'Sullivan
GHC 7.2 compatibility
Bryan O'Sullivan
Finally! A test suite! Admittedly containing just one test, but it's a start.
Bryan O'Sullivan
Correctly handle samples with heavily clustered values (gh-11) If a sample contains many values that are clustered around a single value, this throws off classifyOutliers. Before this change, we could easily end up considering a value as both high and low, thereby counting it in more than one bucket at a time (which should not happen). As a result, we would sometimes report more outliers in a data set than sample values. With this change, every outlier shou…
Bryan O'Sullivan
Use a more suitable wallclock time source on each major platform We now use clock_gettime on most Unix variants, mach_absolute_time on OS X, and QueryPerformanceCounter on Windows. Each of these counters provides much better resolution than their predecessor (getPOSIXTime), and generally at lower cost, too. Nice!
Bryan O'Sullivan
Filter out zero-valued clock differences
Bryan O'Sullivan
Kill off benchmark comparison as a feature I never really reviewed this, never use it, and find the code very obtuse and difficult to understand. This makes it a huge pain to refactor when I want to add new features. It seems the more-correct thing to do is to get rid of this feature and find a better way to implement it that doesn't make the core of criterion less maintainable.
Bryan O'Sullivan
Rework Benchmarkable from a typeclass to a newtype The typeclass gave us no expressivity, but complicated the code somewhat. The new arrangement is quite a bit simpler.
  1. Prev
  2. Next