Split out the caching and source-map-fixing into separate modules.

[jamestalmage]

NYC now does 3 things:

1. Installs an (optionally cacheable) transform via append-transform that captures coverage and source-map data.
2. Remaps istanbul reports using source-map data.
3. Uses spawn-wrap to make all this happen across forked processes.

I think 1 and 2 need to be split out.

[sindresorhus]

👍

[novemberborn]

Remaps istanbul reports using source-map data.

There's https://github.com/SitePen/remap-istanbul but it already does way too many things besides remapping the reports.

Even if 1 and 2 are split out they may still end up coupled to each other. Not sure I see the point, unless they're moved into Istanbul itself maybe.

[jamestalmage]

I think 1 would almost certainly depend on 2 (but not necessarily the other way around).

Not sure I see the point

It would allow all our hard work getting source-maps and caching working to be used in other contexts.

unless they are moved into Istanbul itself maybe

That may be the ideal solution, but I have got to imagine that there is a good reason that source-map support hasn't found it's way to master yet over there.

// @gotwarlost

[novemberborn]

It would allow all our hard work getting source-maps and caching working to be used in other contexts.

Sure, assuming those other contexts discover the modules.

[jamestalmage]

Sure, assuming those other contexts discover the modules.

You take the risk of being ignored every time you type npm publish. That's just life.

For my part, I am really curious how much overhead spawn-wrap and foreground-child add? How much faster would AVA be if we used those split out modules and provided a --coverage option that called them in every forked process? Our babel.js script runs before any user scripts, so we can already provide many of the guarantees that spawn-wrap provides for NYC. Of course, that wouldn't cover users who are going to be forking again from within their test files. In that case - they need to use NYC.

[bcoe]

@jamestalmage it looks like @gotwarlost has been working on his own source-map library as part of the istanbuljs org, could we look at merging efforts?

[coreyfarrell]

Please correct me if I'm wrong but this is resolved by the existence of caching-transform and istanbul-lib-source-maps?

[coreyfarrell]

I think caching is resolved and any further source-map work should be incorporated to istanbul-lib-source-maps.