Author = "Ben Arthur"
Running Pkg.test("Gadfly") evaluates all of the files in
Gadfly/test/testscripts. Any errors or warnings are printed to the REPL. In
addition, the figures that are produced are put into either the devel-output/
or master-output/ sub-directories. Nominally, the former represents the
changes in a pull request while the latter are used for comparison.
Specifically, runtests.jl examines the currently checked out git commit, and
sets the output directory to master-output/ if it is the HEAD of the master
branch or if it is detached. Otherwise, it assumes you are at the tip of a
development branch and saves the figures to devel-output/. After running the
tests on both of these branches, executing compare_examples.jl displays
differences between the new and old figures. This script can dump a diff of
the files to the REPL, open both figures for manual comparison, and/or, for SVG
and PNG files, display a black and white figure highlighting the spatial
location of the differences.
So the automated regression analysis workflow is then as follows:
- In a branch other than master,
- develop your new feature or fix your old bug,
- commit all your changes,
Pkg.test("Gadfly"),- checkout master,
Pkg.testagain,Pkg.add("ArgParse")and, for B&W images, Cairo, Fontconfig, Rsvg, and Images as well,- check for differences with
julia test/compare_examples.jl [--diff] [--two] [--bw] [-h] [filter]. For example,julia test/compare_examples.jl --bw .js.svgwill show black and white images highlighting the differences between the svg test images.