Status of timers enroll/unenroll/active

[piscisaureus]

The timers module exposes some functions (enroll, unenroll, active) that allow for the implementation of efficient idle timers and are used for this purpose by the builtin http module.

I'd like to make some changes to these APIs in a semver-minor patch, but the status of these APIs is currently unclear.
These functions aren't documented and their use outside of core seems very limited.
My proposal is to define them as being private and renaming them to '_enroll', '_unenroll' etc. as part of the next change.

Comments? @iojs/tc

[trevnorris]

If it's in heavy use by user-land then I'm not for changing them. Otherwise +1 on the idea.

[mikeal]

I've been working on a tool that could tell us how many modules in npm use a particular core module, either themselves or through one of their dependencies. It sounds like it would be useful here :)

[bnoordhuis]

FWIW, a quick if non-exhaustive GH code search doesn't turn up any projects that use enroll/unenroll.

[mikeal]

wow... if GH code search finds nothing then I'm pretty confident nobody is actually using this :)

[cjihrig]

I'm also +1 if it's not in use. I'd be surprised if it was. I recently updated enroll() and it was only used like twice in core IIRC.

[mikeal]

I was able to find a few, looks like mostly people using or building polyfills for browserify though https://github.com/search?l=javascript&q=timers.enroll&ref=searchresults&type=Code&utf8=%E2%9C%93

[Fishrock123]

@chrisdickinson maybe you could run this against your packages thing?

[Fishrock123]

@piscisaureus what changes were you planning to do to them?

[Fishrock123]

See: http://logs.libuv.org/io.js/2015-06-24#23:15:56.833

Bert's request was (basically) to make them always act like _unrefActive.

[Trott]

@Fishrock123 Is the plan to first deprecate active(), enroll(), and unenroll() and then make them private at some semver-major point in time? Or something else?

Would it be useful to see if @ChALkeR can pull up some rough usage stats for userland module?

[Fishrock123]

@Trott I think we should probably actually keep them, undocumented.

They are non-problematic and unlikely to ever require significant changes imo.

Closing this, re-open if necessary.