Only one entry per normalized url

[sindresorhus]

There's a lot of duplication in the registry which causes confusion and bloat.

We should only allow unique normalized urls in the registry.

Example: http://sindresorhus.com/bower-components/

Both twitter and bootstrap is registered to the same url. Seen a lot of this with other names too.

[patrickkettner]

@sindresorhus - I believe this is about as fixed as it can be now that the url rewriting is in place, no?

[sheerun]

I listed 1250 packages to normalize: https://gist.github.com/sheerun/cfa608d52bc539422f71

Some are weird like:

  {
    "name": "polymerlabs-yt-video",
    "url": "git://github.com:PolymerLabs/yt-video.git",
    "new": "git://github.com/:PolymerLabs/yt-video.git"
  }

But I guess they modify already broken packages anyway.

Let me know if you see something weird.

[sheerun]

Those are all unusual renames: https://gist.github.com/sheerun/538229188e255b09b2e7

[rayshan]

This may be something missed by https://github.com/bower/registry/pull/82.

@patrickkettner, I see that there are tests for

it('should support ssh urls', function () {
  return assert.equal(normalizeURL('git@github.com:bower/Bower.git'), normalizedURL);
});

it('should support ssh urls with a preceeding slash', function () {
  return assert.equal(normalizeURL('git@github.com:/bower/Bower.git'), normalizedURL);
});

But what happens if the input is git://github.com:PolymerLabs/yt-video.git?

[patrickkettner]

Those aren't valid git repo URLs. They wouldn't pass the new verification, and bower install fails for them currently

[patrickkettner]

To clarify, they are mixing the git protocol with ssh style repo listings

[sheerun]

It's already implemented (normalizing as well)