Send an event at startup with the protocol version and pid - #17873

code lgtm

It looks like there are test failures based on the unit tests not expecting the new, initial event.

Re: updating the protocol docs, doing that after that lands sgtm. We could also (after this PR) move the docs to a markdown file in the repo, and update the wiki to point to that doc. That would allos us to do atomic changes of the protocol and its docs.

It looks like there are test failures based on the unit tests not expecting the new, initial event.

I'll take a look. I wasn't sure how to reliably run all of the tests so I figured my friend Travis would help :)

We could also (after this PR) move the docs to a markdown file in the repo

Yeah, that sounds good to me. I'll take a look once this PR is good to go.

There are still some weird failures on Travis, like this:

05:46 +344 ~1 -3: test/devfs_test.dart: devfs remote cleanup preexisting file system [E]
  Expected: equals ['create test'] unordered
    Actual: [
              'writeFile test .packages',
              'writeFile test lib/foo.txt',
              'create test',
              'destroy test',
              'create test'
            ]
     Which: has too many elements (5 > 1)
  
  package:test                                         expect
  test/src/mocks.dart 384:5                            BasicMock.expectMessages
  test/devfs_test.dart 356:17                          main.<fn>.<fn>

Doesn't seem like something I could've caused, but I'll do some digging next week before landing this.

Re-reading the logs, I think the devfs failures are as a result of a timeout:

Test timed out after 30 seconds

I think the subsequent tests failed because the test that timed out didn't clean up properly. The tests seem to pass locally for me.

I've rebased on master which is forcing the tests to run again, so if they all pass this time I'll land this.