assemble all db repos

[dannycoates]

fixes #54

New layout:

The db-server dep is moved in at the root level and required as a dependency by package.json and the mem bits are now intermixed in bin and lib. This allows everything to be in sync while allowing us to publish fxa-auth-db-server as a module to npm for 3rd party implementations.

Eventually we might want to rename this repo to reflect that its not just mysql, but its no big deal to leave it, and nothing will need to change for fxa-dev or prod. I've already got this running locally with fxa-auth-server with only a few minor changes.

├── CHANGELOG
├── Dockerfile
├── Gruntfile.js
├── LICENSE
├── README.md
├── bin
│   ├── db_patcher.js
│   ├── mem.js
│   └── server.js
├── config
│   ├── config.js
│   └── index.js
├── coverage.html
├── fxa-auth-db-server
│   ├── AUTHORS
│   ├── CHANGELOG
│   ├── CONTRIBUTING.md
│   ├── Gruntfile.js
│   ├── LICENSE
│   ├── README.md
│   ├── coverage.html
│   ├── docs
│   ├── grunttasks
│   ├── index.js
│   ├── lib
│   ├── package.json
│   ├── scripts
│   └── test
├── index.js
├── lib
│   ├── db
│   ├── logging.js
│   ├── notifier.js
│   └── promise.js
├── npm-shrinkwrap.json
├── package.json

Note, this requires npm >= 2.0 for the "file" type dependency.

[jrgm]

Looks good. But looks like you'll need to npm install -g npm@2 in the .travis.yml because the 0.10 build is using npm @1.4.28 unfortunately.

[dannycoates]

Thanks @jrgm, now we're green

[dannycoates]

maybe it would be less weird if instead of the npm dep if we just changed the requires. I was dazzled by the new shiny 'file:' option.

[dannycoates]

maybe it would be less weird

except then we'd need to promote the db-server deps up. clearly i need to stop looking at this. 🍺

[chilts]

Things I've looked at which seem fine:

• npm test in the top level seems fine (including patching the database)
• npm install && npm test in fxa-auth-db-server seems fine
• node bin/mem.js seems fine
• npm start starts a mysql server fine

Random minor queries:

• db_patcher.js is now executable but is only ever run using node (doesn't have a #! interpreter) ... minor, but is it needed?
• I think the tests are running the mysql backend (test/backend/db_tests.js requires '../../lib/db/mysql' same in test/backend/remote.js) but neither are doing the same with the mem backend, is that true? If so, should it do both? This is probably my main concern here. 🌵

Apart from delving into all of the code (which just looks like a re-arrangement with probable changes to requires) then this all looks fine to me! :) r+ with nit regarding backend tests. Totally bogus @dannycoates!

[rfk]

Will this .travis.yml in a random subdirectory actually get run?

[dannycoates]

no

[rfk]

I think the tests are running the mysql backend (test/backend/db_tests.js requires '../../lib/db/mysql'
same in test/backend/remote.js) but neither are doing the same with the mem backend,

If true, we need to figure out a way to run both sets of tests under travis

[rfk]

IIUC the next step here is for danny to comment on chilts' queries above, then we can rebase for a merge.

[rfk]

I went ahead and rebased this and merged in train-42 from the other repos. Let's get it merged and we can file follow-up bugs for further cleanup. Merging r=@chilts

[dannycoates]

Eh, sorry I haven't looped back to this yet. @rfk, thanks for the merge anyway. I see @philbooth took the followups, lmk if I can help with anything :)