Honor excludes, but use faster algorithm about grunt-jslint

[smikes]

This is issue 42, branch is smikes/grunt-jslint exclude-On-42 (will come in as a separate pull request because I did this badly...)

[stephenmathieson]

Should this be closed?

[smikes]

Let's wait to see if @sterpe weighs in in the next couple days.

On Mar 20, 2014, at 10:28 AM, Stephen Mathieson [email protected]
wrote:

Should this be closed?

Reply to this email directly or view it on
GitHubhttps://github.com//issues/42#issuecomment-38188867
.

[sterpe]

@smikes
Either way they're both O(n) algorithms so should be better in general. I think the algorithm cribbed from JSHint code base is probably faster in situations where someone has a lot of files and is using very few very broad globs, i.e,

['src/**/*.js', '!**/node_modules/**']

since it doesn't actually try to expand all the files in the (potentially huge) exclusion glob.

If the globs are extremely specific, I imagine the map is faster.

The one problem with the map approach is because it doesn't regex the exclusion glob against an fp it's subject to issues of equivalent paths having inequivalent strings -- i.e.,

['./foo/**/*.js', '!**/node_modules/**'] 

in this case, I think that

**/node_modules/** 

is going to expand to:

foo/node_modules/... 

so all of the exclusions may be ignored (but we should check this I think).
easy fix is

[..., '!./foo/**/node_modules/**'] 

but an issue to be aware of.

[sterpe]

FWIW, I guess the issue with #41 is that we didn't path.resolve the exclude paths so the minimatch failed.

I'm fine with whatever approach. Thanks @smikes @stephenmathieson