doc: Remove 'dnt_helper.js'

[ghost]

This file is totally useless, because it's an inner-used helper that is for publishment of doc files.

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• documentation is changed or added
• commit message follows [commit guidelines]

[Trott]

I wonder if the file can be removed entirely and any calls to _dntEnabled() replaced with navigator.doNotTrack || window.doNotTrack. (The window check is for IE 11. If you want to be REALLY accommodating and account for IE 9 and IE 10, throw in a || navigator.msDoNotTrack.)

This file seems to exist mostly for operating systems and browsers we probably don't need to take particular care in accommodating. (Windows NT 6.1??!! Firefox 31??!!) To get an idea of just how broadly the Do Not Track API is supported on current browsers: https://caniuse.com/#search=dnt and click "Show All":

Note that even for browsers that show red in that chart, it's not a problem. It just means the browsers don't support Do Not Track, which just means it will never be enabled on those browsers and that's OK. We'll just get the falsy undefined for those browsers. If the user can't set Do Not Track in their browser, then we aren't failing to respect Do Not Track.

[Trott]

@nodejs/website on my previous comment ^^^^^^^

[ChALkeR]

This change will disable analytics on IE versions that break the spec and default DNT to enabled (previously the helper ignored those).

That should be fine, I guess.

---

I have another (unrelated here) question about analytics though. I remember claims that it will improve docs back when it was added.
Are there any issues, comments or PRs that rely on data from the analytics? Has it proven to be useful at all?

[silverwind]

Windows NT 6.1

This is Windows 7 which still has a considerable userbase.

Are there any issues, comments or PRs that rely on data from the analytics

The question is: Who has access to this analytics account and are they actively looking at the data? If not, I suggest dropping it entirely.

[fhemberger]

Stats for last month (65% of the Visitors use any version of Windows):

Windows 10: 70.0%
Windows 8:   5.5%
Windows 7:  24.0%
older:     < 1.0% 

Browsers used by Windows visitors:

Chrome:           81.5%
Firefox:           9.5%
Edge:              4.0%
Internet Explorer: 3.0%

@silverwind Anything else you need?

[ChALkeR]

See also previous discussion in #6601.
/cc @phillipj

[silverwind]

@fhemberger thanks, no it's fine to keep if we use this data. The current PR should cover IE9 and up, seems good enough and a nice simplification over the previous checks.

[BridgeAR]

CI https://ci.nodejs.org/job/node-test-pull-request-lite-pipeline/742/

[addaleax]

Landed in ec75ba6

[ghost]

Thanks!