Follow symlinks

[valeriangalliat]

• Removed getStats function.
• Added a resolveSymlinks function that exposes a stat property on file object as a side effect (making getStats obsolete).
• Added tests.
• Fixes Symlinked directories cause EISDIR error from vinyl-fs.src() #39.

[coveralls]

Coverage decreased (-0.86%) to 96.62% when pulling a4ea945 on valeriangalliat:symlinks into 041de1b on wearefractal:master.

[coveralls]

Coverage decreased (-0.86%) to 96.62% when pulling a4ea945 on valeriangalliat:symlinks into 041de1b on wearefractal:master.

[yocontra]

Should this be the default behavior or behind a flag?

Also code coverage decreased from 100%

[valeriangalliat]

Should this be the default behavior or behind a flag?

I'd prefer to leave this decision to someone that knows vinyl-fs better than me. 😄

Also code coverage decreased from 100%

Before:

if (stat) {
  file.stat = stat;
}
cb(err, file);

After:

if (err) { // not covered
  return cb(err);
}
file.stat = stat;
cb(null, file);

That's roughly the kind of changes I had to make when switching from getStats to resolveSymlinks, because while the first format allows to increase code coverage without having to explicitely test the error case, it makes the code less readable, mostly when you have nested calls like this.

I have even no idea what I have to pass to vfs.src to make fs.lstat and path.realpath "return" an error… if I pass an unexisting file, it will be handled at the glob level and will never reach the resolveSymlinks function. Any idea?

[yocontra]

Hmm not really sure how to test those. I guess you would have to mock the fs module and have it return errors and make sure they bubble up. Seems like overkill though IMO - thanks for the PR

[yocontra]

Will be published as a part of the 4.0 release when it is ready. 4.0 = master branch right now