src: begin restructuring node.cc for better maintainability

[jasnell]

Begins separating logical chunks of node.cc out into separate cc/h files for the purpose
of making these easier to maintain and making node.cc less of a monolithic jumble of code.

There is quite a bit in node.cc that can be divided into logical chunks and this PR only handles two of those. Before going too much further with it tho, I'd like to make sure folks are on board with this.

The eventual goal is to refactor the bootstrap model for node.js with an eye towards improving start up times, supporting snapshots, and generally making the bootstrap more maintainable. However, because there is so much bundled into node.cc, it is difficult to refactor and reviewing changes to make sure nothing breaks is quite challenging.

And yes, the idea would be to backport these changes to LTS lines as well.

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• commit message follows commit guidelines

[jasnell]

CI: https://ci.nodejs.org/job/node-test-pull-request/14913/

[mhdawson]

To help in reviewing, does this change code or simply move to a new file. If there is code that had to be changed in more than a trivial way and it is not the bulk of the new files it would be useful to identify those portions.

[jasnell]

The first and third commits simply move things around. There are no implementation changes.
The second commit changes a struct to a class to fix a weird linting issue.

[addaleax]

LGTM, assuming you are volunteering to do backports :)

[jasnell]

Yes, I'll do the backports for 8 and 6

[TimothyGu]

Code changes LGTM. Do we need to make the new files inherit the license header in node.cc?

[jasnell]

@TimothyGu ... no, the requirement for the license header was only maintaining it in the original files. New files do not need it. I know it's weird, but that's the agreement we came to.

[jasnell]

@addaleax ... looking at the backport to 8... it's going to be quite a bit more involved because of the DomainEnter/DomainExit and related functions. I'm not entirely convinced we should backport this particular change to 8 and really don't think we should backport to 6. Thoughts?

[addaleax]

@jasnell Makes sense … but we can backport the exceptions stuff, right?

[jasnell]

We ought to be able to do at least part of it, yeah, would just need to be selective about which pieces.

[jasnell]

Landed in fbd1c49, c68e6e1, and 9349e15.
backport PRs will come next week.

[Trott]

Is there a reason c68e6e1 wasn't squashed into fbd1c49? Or was it just an oversight? Landing them separately means that fbd1c49 is a broken commit on master. 😱

[richardlau]

This had broken addons because the new headers referenced in node.h are not packaged in the headers tarball that node-gyp downloads: #20921