Catch-all issue

[twolfson]

We have a lot of development work left but the initial proof of concept is out in the wild. Here's what's left on our plate:

• [x] TODO: Add JSON reporter to Gemini or build out multi-image-assert
  • Definitely add an assert tool or an --no-server flag so we can exit early
• [x] TODO: Simplest API will be gemini-test --reporter json || multi-image-mergetool --preset gemini (loads Gemini config, finds screens based on that, (skips initial assertion), starts server + opens browser window)
  • We are settling on making this tool generic. If someone wants anything more complex than gemini-test --reporter html || multi-image-mergetool --loader gemini (e.g. a fully integrated UI like gemini-gui), then they can write it
• [x] TODO: Move to statically saved images instead of Gemini generated ones
• [x] TODO: Test out via karma and Sauce Labs/BrowserStack?
• [x] TODO: Move from Jade to HTML so we can have gh-pages demo Unnecessary, we can use Jade for demo page as well
• [x] TODO: Add gh-pages demonstration of tool (with mock server responses)
• [x] TODO: Clean up scripts and dependencies
• [x] TODO: Include performance tests in final test suite
• [x] TODO: Document CLI flags (--help)
• [x] TODO: Add --preset gemini flag support
• [x] TODO: Remove external dependencies (e.g. rawgit, jQuery CDN)
• [x] TODO: Standardize on imgSet vs imageSet
• [x] TODO: Reset dev via: rm -r gemini gemini-report; cp -r gemini-report-bak/* .
• [x] TODO: Move to Bootstrap website for our development and demo images
• [x] TODO: Mockups reference https://gist.github.com/twolfson/c4236abadeada82e2686c940fb23341d
  • [ ] TODO: Add a logo?
• [x] TODO: Testing strategy -- Karma for browser code, with plenty of isolation for things like overlay, button scenarios, handling HTTP request (fake responses via sinon or a fixed-server). Mocha for server tests -- maybe selenium for integration test but think it's overkill. Maybe use something from server tests as fixtures for browser tests so we have contract agreement on HTTP interface
  • [x] Test CLI (or at least test it with mocked calls)
  • [x] Test ImageSet
    • Not tested directly but covered all interfaces via CLI and server
  • [x] Test server HTTP interface
  • [x] Test browser UI
• [ ] TODO: Add CLI logging on image accept/update, its result, and images remaining
• [ ] TODO: Add fixed nav bar which allows jumping to previous/next conflict
• [x] TODO: Move to localized CSS as well (i.e. no rawgit)
• [x] TODO: Explore moving from domo to hyperscript/hyperscript-helpers
• [x] TODO: Enable our asserts
• [x] TODO: Remove test/test-files/gemini as we no longer have setup for it/use it
• [ ] TODO: Add CSRF support
• [x] TODO: Get rid of 404 warnings in Karma tests by adding /base/ images instead of data-uri Repaired by moving to imgUrl for all images and ditching prefix (we were going to have similar problem as we would need a /base prefix and no encoding)
• [x] TODO: Add visual tests (mostly gathered in browser suite)
  • [x] Gather screenshots
  • [x] Generate images for current, diff, and ref (thinking 3x3 checkerboard, 1 dot center, and diff). Maybe square border and vertical line for other?
  • [x] Integrate screenshot validation into tests (hopefully using our tool, requires moving off of commander.js as it lacks support for variadic arguments in options, i.e. --current-images [images...])
  • [x] Explore how to integrate screenshots into CI (considering hashing as a low cost option for now, previous imgur integration was meh)
• [x] TODO: Make sure we test all our expected CSS (browser, index) and scripts are loaded in server page (not sure what I was thinking here...)
• [x] TODO: Update test-files/README.md to use multi-image-mergetool instead of one-off script (requires --diff-images support on CLI)
• [x] TODO: Update test-files/README.md to use multi-image-assert (currently listens and needs Ctrl+C)
• [x] TODO: Be sure to test --current-images/--ref-images not existing with no --loader
• [x] TODO: Be sure to test --loader with Gemini
• [x] TODO: Be sure to test imageSet#options.diffImage support properly works
• [ ] TODO: Add new features to overlay such as custom HTML handles for dragging and resizing (corners, edges)
• [x] TODO: Move ImageSet in DOM to class with singleton model for quick look up by id (will need to add reset method for tests but that's fine)
• [ ] TODO: For overlay highlighting on similar images, create a canvas with rgba and clipped area, clone 3x, and position: absolute over image
  • We prefer this over alternative of fading original image as that requires duplicating images due to preserving original updated ref (for saving) and importing original current into a canvas for fading
• [x] TODO: Add smooth scrolling to jump to similar images We have decided against this as we can't predict if overlay found all the images the user desired =/ (e.g. found 2 when wanted 20)
• [x] TODO: Fix overlay being visible after image set collapse
• [x] TODO: Add support for missing images in ref-images (i.e. current image is new)
  • Should be achievable via an isNew property
• [ ] TODO: Remove our TODOs and console.logs
• [ ] TODO: Complete all skip tests
• [ ] Add support for remaining loaders
• [x] Complete overlay tests
• [ ] See test plans at bottom of one of the test files
• [ ] Determine why overlay jumps on image set collapse (might be related to #3)
• [ ] Verify we don't search approved images for "Find similar images"
• [ ] Remove querySelector calls from browser directory, there shouldn't be any need with this.saveEl
• [x] Explore placing mutation updates inside of render with previousState, newState as parameters (should handle loading, cachebusting images (save URLs on state), image equality, results count -- should be closer to title el)
• [ ] Add lightbox expand support a la Imgur for shrunken images/canvases
• [ ] Overlay loses position on window width adjustment (also look at height adjustment)

[twolfson]

We will be using the following screenshot (test/browser/actual-screenshots/find-similar-matching.png) for our README. Please update screenshot images in this thread and update the link in the README. We don't want to bloat our repo with screenshot images (it's about 30kb crushed).

We initially had a setup for this but we didn't like the bloat:

"test-browser": "npm run build && npm run test-karma-single && npm run test-copy-docs-screenshot",
"test-copy-docs-screenshot": "if which pngcrush &> /dev/null; then src_file=\"test/browser/actual-screenshots/find-similar-matching.png\"; test -f \"$src_file\" && pngcrush \"$src_file\" docs/screenshot.png; fi",

[twolfson]

Proof of concept test for a base64 image in GitHub issues:

Doesn't work =(

[twolfson]

According to blog posts, we can't submit email attachments either =( It also looks like email thread ids are random UUIDs =/

https://github.com/blog/811-reply-to-comments-from-email

[twolfson]

Latest screenshot for README:

[twolfson]

Latest screenshot: