package.json - Issue 29860555: Issue 6717 - Part 2: run qunit in headless firefox

Side by Side Diff: package.json

Issue 29860555: Issue 6717 - Part 2: run qunit in headless firefox (Closed) Base URL: https://hg.adblockplus.org/adblockpluschrome/file/3270e924ba9f

Patch Set: Created Aug. 23, 2018, 3:30 p.m.

Left:
Right:

Use n/p to move between diff chunks; N/P to move between comments.

Jump to:

View unified diff | Download patch

OLD	NEW
1 {	1 {

2 "name": "adblockpluschrome",	2 "name": "adblockpluschrome",

3 "repository": "https://hg.adblockplus.org/adblockpluschrome",	3 "repository": "https://hg.adblockplus.org/adblockpluschrome",

4 "license": "GPL-3.0",	4 "license": "GPL-3.0",

5 "dependencies": {},	5 "dependencies": {},

6 "devDependencies": {	6 "devDependencies": {

	7 "//": [

	8 "'dmg', 'extract-zip' and 'ncp' are needed by the module",

	9 "adblockpluscore/test/download_firefox.js."

	10 ],

	11 "dmg": "^0.1.0",

7 "eslint": "^5.1.0",	12 "eslint": "^5.1.0",

8 "eslint-config-eyeo": "^2.1.0"	13 "eslint-config-eyeo": "^2.1.0",

	14 "extract-zip": "^1.6.5",

	15 "geckodriver": "^1.9.0",

	16 "ncp": "^2.0.0",

	17 "nodeunit": "^0.9.1",

	18 "selenium-webdriver": "^3.6.0"

9 },	19 },

10 "scripts": {	20 "scripts": {

11 "lint": "eslint *.js lib/ qunit/ ext/",	21 "lint": "eslint *.js lib/ qunit/ ext/ test/",

	22 "run_tests": "nodeunit test/*",

	23 "test": "_run () { python build.py devenv -t $1; npm run run_tests ; } ; _ru n",
	Sebastian Noack 2018/08/23 16:38:44 I'm wondering why do we need to define a sheel scr I'm wondering why do we need to define a sheel script function just to call it right after? Why can't we inline it? Also why make the target for build.py configurable? The tests will run on Firefox regardless. So passing in anything other than "gecko" seems pointless. tlucas 2018/08/23 18:56:02 Making it configurable was the reason to do this, Show quoted text On 2018/08/23 16:38:44, Sebastian Noack wrote: > I'm wondering why do we need to define a sheel script function just to call it > right after? Why can't we inline it? > > Also why make the target for build.py configurable? The tests will run on > Firefox regardless. So passing in anything other than "gecko" seems pointless. Making it configurable was the reason to do this, yes - the argument ("gecko" in our case) would have been passed in some weird way to "npm run run_tests" instead. For "why configurable" -> we might consider this premature optimization, but already given that Edge (and likely Chrome too) will be build on a different machine (and in parallel), this is meant to "build" and "test" for whatever platform is handled by the current CI - job. Then again, this strongly supports adding an abstract test_runner.js right from the start. Please tell me what you think. Sebastian Noack 2018/08/23 21:54:20 It seems we have to touch the CI configuration aga Show quoted text On 2018/08/23 18:56:02, tlucas wrote: > On 2018/08/23 16:38:44, Sebastian Noack wrote: > > I'm wondering why do we need to define a sheel script function just to call it > > right after? Why can't we inline it? > > > > Also why make the target for build.py configurable? The tests will run on > > Firefox regardless. So passing in anything other than "gecko" seems pointless. > > Making it configurable was the reason to do this, yes - the argument ("gecko" in > our case) would have been passed in some weird way to "npm run run_tests" > instead. > > For "why configurable" -> we might consider this premature optimization, but > already given that Edge (and likely Chrome too) will be build on a different > machine (and in parallel), this is meant to "build" and "test" for whatever > platform is handled by the current CI - job. Then again, this strongly supports > adding an abstract test_runner.js right from the start. > > Please tell me what you think. It seems we have to touch the CI configuration again anyway, once support for other browsers have been added, since if the CI (or yourself) would already try to run "npm test -- chrome" it will potentially explode, since with the current implementation this would essentially run a build for Chrome on Firefox. So IMO we could as well just hard-code gecko here for the time being, making sure that always the correct devenv is built for the only browser supported so far. tlucas 2018/08/24 10:25:08 Done. Show quoted text On 2018/08/23 21:54:20, Sebastian Noack wrote: > On 2018/08/23 18:56:02, tlucas wrote: > > On 2018/08/23 16:38:44, Sebastian Noack wrote: > > > I'm wondering why do we need to define a sheel script function just to call > it > > > right after? Why can't we inline it? > > > > > > Also why make the target for build.py configurable? The tests will run on > > > Firefox regardless. So passing in anything other than "gecko" seems > pointless. > > > > Making it configurable was the reason to do this, yes - the argument ("gecko" > in > > our case) would have been passed in some weird way to "npm run run_tests" > > instead. > > > > For "why configurable" -> we might consider this premature optimization, but > > already given that Edge (and likely Chrome too) will be build on a different > > machine (and in parallel), this is meant to "build" and "test" for whatever > > platform is handled by the current CI - job. Then again, this strongly > supports > > adding an abstract test_runner.js right from the start. > > > > Please tell me what you think. > > It seems we have to touch the CI configuration again anyway, once support for > other browsers have been added, since if the CI (or yourself) would already try > to run "npm test -- chrome" it will potentially explode, since with the current > implementation this would essentially run a build for Chrome on Firefox. So IMO > we could as well just hard-code gecko here for the time being, making sure that > always the correct devenv is built for the only browser supported so far. Done.
12 "posttest": "npm run lint"	24 "posttest": "npm run lint"

13 }	25 }

14 }	26 }

OLD	NEW

« no previous file with comments | « dependencies ('k') | test/firefox.js » ('j') | test/firefox.js » ('J')