Make continuous deployment safe by comparing before and after webpage screenshots for each release. Depicted shows when any visual, perceptual differences are found. This is the ultimate, automated end-to-end test.
Renames flags to be not-phantomjs-oriented, now they look like --capture-
Adds capture.py which can be configured by release to use command_executor and desired_capabilities, as supported by Browserstack.
capture.py is in its infancy and still lacks support for a lot of the other capture flags, documented as TODOs in capture.py. But I've tested this on an instance and it works nicely - even when Browserstack's queue gets overloaded, dpxdt's worker queue will retry until it works - which is sweet.
For the record, Firefox on Browserstack has by far the best screenshots. iPhone simulator works too which makes this really compelling as compared to phantomjs which seriously lags behind in support for so many good things in CSS that are now common (flexbox, vendor prefixes, etc..).
--capture-capture.pywhich can be configured by release to usecommand_executoranddesired_capabilities, as supported by Browserstack.capture.pyis in its infancy and still lacks support for a lot of the other capture flags, documented as TODOs incapture.py. But I've tested this on an instance and it works nicely - even when Browserstack's queue gets overloaded, dpxdt's worker queue will retry until it works - which is sweet.For the record, Firefox on Browserstack has by far the best screenshots. iPhone simulator works too which makes this really compelling as compared to phantomjs which seriously lags behind in support for so many good things in CSS that are now common (flexbox, vendor prefixes, etc..).