Issue #4680

[sebastienroul]

fix(location,route): Breaks with file:// protocol in Chrome

FF and IE implementations considere the drive letter as "host" and remove it from pathname
CHROME implementation let it as part of the pathname
We should remove the drive letter from the pathname to let routing work.

Closes #4680

[petebacondarwin]

@sebastienroul - can you ensure that you have signed the CLA? Thanks.
Also you would improve the chances of getting your PR merged quickly if you read and follow the contribution guidelines: https://github.com/angular/angular.js/blob/master/CONTRIBUTING.md

[sebastienroul]

@petebacondarwin : Signed yesterday... Waiting for response :)

[jeffbcross]

Hi @sebastienroul , would you mind changing your commit messages and PR description to be consistent with our git commit guidelines? https://github.com/angular/angular.js/blob/master/CONTRIBUTING.md#git-commit-guidelines

We use commit messages to auto-generate changelogs.

It'd also be helpful if you could write a failing unit test (which passes with your implementation).

[sebastienroul]

Hi @jeffbcross : OK for the commit message : I've update it. Should I do this for all my commit messages or only for the first of the pull ?
2 another questions (sorry, just born yesterday :) ):

• My Travis build failed.. but not lot of explanations why "Safari 6.0.5 disconnect" : https://travis-ci.org/angular/angular.js/builds/13830187
• Very hard to write a test cause it only append when opened document come from the file system: Any idea how to simulate this to get a test in isolation ? (I wrote a small case in the issue : ngRoute breaks with file:// protocol in 1.2.0-rc.3 #4680 (comment)

In all thanks for your patience !

[caitp]

@sebastienroul you can squash these commits and correct your commit message according to the contributing guide with a rebase.

This travis failure isn't a big deal, you can trigger a rebuild by re-pushing to github.

As for writing tests, any change is going to need at least one test case to assist in preventing future regressions. I'm happy to assist in writing tests, or reviewing the tests you write, to ensure that they're as meaningful as possible, and prevent them from breaking as much as possible. Happy to help with this, but to start with you should write a test which verifies that your own particular use case is served.

[jeffbcross]

No worries ;)

1. All commit messages should follow the format in the guidelines
2. Don't worry about Travis; it's flaky right now
3. Don't sweat the test; I can write one and amend it to the commit

[jeffbcross]

Ahh yes, what @caitp said... a squash of commits would be helpful.

[jeffbcross]

@sebastienroul I'm reviewing this PR now, and I'll go ahead and squash the commits and change the message.

[sebastienroul]

Thanks for all ! continuous improvement :)
Ask me if any question on the problem itself.

[jeffbcross]

@sebastienroul, I amended a test and some implementation changes to your commit, and re-opened it under this PR: #4928