Maybe this fixes it in eslint 4?
eslint/eslint#8706

Hmm not sure. I'm not sure why caching should fix it, but I know almost nothing about eslint internals, so I could very well be wrong.

AFAIK standard doesn't support eslint 4. I am right?

@ematipico Correct. I think what @Flet was hinting at was if the solution to this might come automatically when at some point standard is upgraded to 4.x

@watson could you try with standard@beta? It uses the latest version of eslint. At least we could determine if its fixed? :)

@Flet doesn't seem to fix it unfortunately:

hmmm hang on... might have spoke too soon. Let me just check something

Nah, I wanted to try to also ignore the folder inside package.json, but no dice. Here you can also see the time it takes if I delete the folder:

Aww, was worth a shot :(

I'm having this same issue. Fix is traversing ignored paths and complaining with a exit code 1

Can confirm, I have the same issue with a very-very-very large (almost cyclic) examples/node_modules ignored directory in testdouble.js (still reproducible at this commit). All major versions of Node.js literally run out of heapspace attempting to scan it.

The issue seems related to deglob. Posting more info from @searls:

LOL amazing issue report

Nice thank you for digging @serls :)

I took at stab at this and found that this is a combination of a few issues.

As @searls pointed out, the issue is inside deglob.

1. Diving further, the time is spend inside the glob module: https://github.com/standard/deglob/blob/bce559a275faf286395c52912a82c601a6740f2d/index.js#L23-L27
2. The ignore option passed to the glob module does not contain patterns found in the .gitignore file, so files only ignored here will still be processed by glob
3. Even if the lines from the .gitignore file were passed to the ignore option, these would have to end with /** to ignore whole directories. This is inconsistent with how .gitignore works and with what's documented in the README

Suggested solution:

1. Document that you need to append /** to directories you wish to ignore in the package.json standard.ignore array
2. Convert the patterns from .gitignore to patterns compatible with the glob module and add them to the list of ignored glob patterns

@feross what do you think of this?

My preferred solution would be to update deglob to behave like the glob module or .gitignore when it comes to foo/**/* style patterns.

I realize it could be seen a breaking change, but one person's breaking change is another's patch fix! Based on the popularity of using **/* to mean "all files, recursively" I can't imagine anyone is intentionally passing deglob ignore rules of that type with an expectation they will not be ignored recursively.

Is that on the table?

I'm removing deglob for standard 16. So this issue will be fixed for free!

Woohoo! Thanks @feross!

This will be fixed in standard 16