The most egregious to me: prettier/prettier#5588

This is super exciting! Thanks for documenting all the issues you're facing during the migration.

As a quick workaround of the html template issue, I believe that you could call .trim() on the string and that would fix the issues.

I'm curious, what tooling is using the /*OK*/ annotation?

/*OK*/ is our own annotation system. Originally, we used a primitive regex for .foo, and reported warnings. So, putting any comment between the . and foo made the regex not match.

Now, we use eslint and check the leadingComments on the property. But we've stuck with /*OK*/ comments.

We have a new Prettier release with bug fixes at https://prettier.io/blog/2019/04/12/1.17.0.html. (Yay!) I synced to master, re-ran the auto-fixer, and pushed a new commit at 1d990a5.

Current status:

1. Quoted object keys are now being preserved (Respect quoted object keys, as unquoting breaks aggressive minifiers like Closure Compiler prettier/prettier#4327) ✔️
2. Inline /*OK*/ tags are still being moved across parentheses (link, Inline single-line comments next to parentheses are being moved incorrectly prettier/prettier#5973) ✖️
3. Typecast parentheses are still being moved incorrectly (link, JSDoc type cast parentheses is removed  prettier/prettier#4799 (comment)) ✖️
4. HTML template strings are still being reformatted (link, Disable embedded language formatting by default and make it configurable prettier/prettier#5588 (comment)) ✖️

There may be ways to work around 2 and 4, but I think 3 is still a dealbreaker.

/to @cramforce @jridgewell @alanorozco @aghassemi for comment
/cc @vjeux for visibility

I ran some checks on the new code.

• Item 2 (inline /*OK*/ tags) affects just 4 or 5 function calls, so I think we can work around it. See the output from gulp presubmit
• Item 3 (typecast parentheses) is a blocker. The output from gulp check-types detects only a subset of incorrect typecasts.
• Item 4 (HTML template strings) can be worked around by using .trim(), but we'd need a mechanism to enforce its usage.

4 need not be fixed for us to adopt prettier, our code will work just fine. My dislike for it is only because it is an actual code change, not a formatting change.

I think only 3 needs to be fixed before we can merge this.

Update: We have a new Prettier release 1.17.1. With this, none of the issues mentioned in #21212 (comment) are blockers any more.

This PR has been updated with the latest version, and is now ready for review. See PR description for details on what has changed.

This PR is now ready for a full review. Some notes:

• The overwhelming majority of changes come from the commit titled First pass of prettier across entire codebase (d34188b), which contains the automatic fixes applied by prettier. This commit can be skimmed through, since we've eliminated breaking changes in the above comments.
• The rest of the commits contain manual changes made by me (see PR description), and should be reviewed as usual.

Tagging a few reviewers to distribute the load. Please post a comment if you see any objectionable changes in the code you own: @cramforce @aghassemi @choumx @newmuis @lannka @Gregable @jpettitt @kristoferbaxter @rsimha

Once this is approved for merge, I will rebase the entire PR on the latest HEAD, rerun prettier, and make sure Travis is green before merging. (I'll likely do this late on Friday evening.)

/cc @gmajoulet @Enriqe can you take a look for stories while I'm out?

@newmuis sure!