tools, test: update ESLint to 2.7.0

[silverwind]

Checklist

• tests and code linting passes
• the commit message follows commit guidelines

Description of change

Updated ESLint from 2.1.0 to 2.7.0 and fixed issues that arose.

Would like some opinions on the prefer-const changes. The rule complains that the variables weren't reassigned and could be made const, but that turned out to be problematic in a few cases where the variable was reassigned from undefined and couldn't be made const in the place of assignment because const doesn't hoist and the variables were accessed earlier.

I'm leaning towards a rule bug in these cases and disabled the rule in the affected tests.

[silverwind]

Here, prefer-const was complaining about server.

[mscdex]

Couldn't those few tests just be modified to use const instead of let?

[silverwind]

Couldn't those few tests just be modified to use const instead of let?

Not without altering behaviour, I think. Take this example:

let foo;

setImmediate(function() {
  console.log(foo);
});

foo = process.hrtime();

Using const, that would be:

const foo = process.hrtime();

setImmediate(function() {
  console.log(foo);
});

foo would be off by one tick, right?

[mscdex]

For example, I meant change this:

let client;

// ...

server = net.createServer(function(c) {

to this:

// ...

const server = net.createServer(function(c) {

and this:

let before;

// ...

before = process.hrtime();

to this:

// ...

const before = process.hrtime();

[silverwind]

Oh, you're right. Looks like your suggestion works and I've been misunderstanding hoisting:

$ node --use-strict -e 'let foo; setImmediate(function() {console.log(foo)}); foo = 1'
1
$ node --use-strict -e 'setImmediate(function() {console.log(bar)}); const bar = 1'
1

[silverwind]

Updated with @mscdex's suggestion.

CI: https://ci.nodejs.org/job/node-test-pull-request/2234/

[silverwind]

(re-pushed latest commit for the gpg signature)

[silverwind]

Looks line one unrelated failure on the CI. Linting took 1min 5sec which seems to be similar to previous version, maybe a tad slower.

[mscdex]

LGTM

Linting will get faster if #5638 eventually lands ;-)

[Fishrock123]

kinda annoying but LGTM

[Trott]

This change adds Intl as a valid global from the linter's point of view. \o/ LGTM

[thefourtheye]

LGTM

[silverwind]

This change adds Intl

That's not in yet, as it was added in globals@9.4.0, but ESLint has globals@9.2.0.

[Trott]

That's not in yet, as it was added in globals@9.4.0, but ESLint has globals@9.2.0.

It's in there. ESLint specifies globals 9.2.0 or above (as long as it's 9.x.x). This PR has 9.4.0 installed. CELEBRATE!

[silverwind]

@Trott indeed. Today I learned that semver carets also match on minor (thought it was patch only):

> require("semver").satisfies("9.4.0", "^9.2.0")
true

[silverwind]

Thanks for reviews, landed in 2f6ff1b and ffe5c83.

[MylesBorins]

@silverwind lts?