Collect timeout feature

[DylanC]

Putting this up early to have eyes on it. Node Clinic will have a new feature to set a timed delay before we collect the data.

Relevant PR:
clinicjs/node-clinic#208

For 0x the delay won't hold up the process but rather filter the data depending on the delay value passed in. Open to feedback and further changes/suggestions.

[goto-bus-stop]

two nits. we should also add a test but we can discuss that in slack a bit later.

[goto-bus-stop]

(This functionality is tested in clinicjs/node-clinic-flame#107. Flame is a bit better set up for testing atm 😬 )

@DylanC we need a line of documentation for this option in docs/api.md. I think there's also another call to the v8LogToTicks() function in index.js that needs to be updated for the new call signature.

[DylanC]

@goto-bus-stop - I have both of those updates pushed into the branch.

[mcollina]

This needs unit tests

[DylanC]

@goto-bus-stop - Should I wait until we figure out the tests for Flame first? Seems testing this would have the same issue?

[ErisDS]

Wondering if there's anyone around to help me revive this PR?

I have a rebased version of this running here with added usage info + a minor tweak:

https://github.com/TryGhost/0x

It is working & useful - it sort of solves #129 as although you can't explicitly say when to start profiling, you can skip the startup process of a script and only measure e.g. when firing requests.

I did look at adding the requested unit tests, but as far as I can tell the code that this affects in v8-log-to-ticks is currently untested, meaning adding a test for this behaviour is a fairly big task, and I don't really know where to start.

I don't really want to run a fork, ideally I'd like to get this populated through flame & clinic as was the intention of the original author.

[mcollina]

would you like to do a fresh PR?

[ErisDS]

Done: #239 :)

[ErisDS]

Heads up there's nothing in here that wasn't in #239, and there for this can be closed 😺