If a function receives a callback argument then it should not return a promise
if the caller's callback will be invoked. Both invoking a callback and
returning a promise can lead to at best confusion (in code review and at
runtime) and at worst non-deterministic behavior, such as race
conditions. Also, a caller supplying a callback may not handle a returned
promise, leading to unhandled rejection errors.
Refactor all readily identified functions where a callback argument can be
supplied but the function returns a promise regardless. Make use of
`callbackify` and `promisify` where it made sense to do so during the
refactoring. Some callsites of the revised functions may have been accidentally
overlooked and still need to be updated. Some functions that take callback
arguments may execute them synchronously, at odds with control flow of a
returned promise (if a callback wasn't supplied). Such cases should be
identified and fixed so that asynchronous behavior is fully consistent whether
the caller supplies a callback or receives a promise.
Make sure promises that pass control flow to a callback ignore rejections,
since those should be handled by the callback.
Don't return promise instances unnecessarily from async functions (since they
always return promises) and change some functions that return promises to async
functions (where it's simple to do so).
Whisper was using an ad hoc promise-like `messageEvents` object. However, that
object behaved more like an observable, since promises either resolve or
reject, and only do so one time. `messageEvents` was also intertwined with
callbacks. Replace `messageEvents` with RxJS Observable. `listenTo` now returns
Observable instances and callers can subscribe to them.
`Blockchain.connect` of embarkjs could suffer from a race condition where tasks
associated with `execWhenReady` might be ongoing when `connect`'s returned
promise resolves/rejects (or a caller supplied callback fires). Attempt to
ensure that returned-promise / supplied-callback control flow proceeds only
after `execWhenReady` tasks have finished. The control flow involved
is... rather involved, and it could use some further review and refactoring.
Bump webpack and the hard-source-plugin for webpack.
[util]: https://www.npmjs.com/package/util
[PR 1318][PR1318] introduces a monorepo member that's used as a DApp
dependency, but the present arrangement whereby `test_dapps/` is an embedded
monorepo makes it difficult to develop and test such packages in tandem with
changes to `test_dapps/packages/*`.
Reorganize `test_dapps/*` as members of the root monorepo and make various
adjustments accordingly. This makes it easy to develop test dapps and any
packages they depend on simultaneously, but we lose the CI/QA arrangement where
test dapps are run with an embark installed from fresh tarballs. That
arrangement, which is quite worthwhile as a means to detect problems arising
from transitive dependencies as soon as possible, will be re-introduced in
another PR, possibly involving an auxiliary repository such as
embark-framework/dapp-bin.
Since the `package.json` `"test"` scripts of `test_dapps/*` are now kicked off
as part of `yarn test` in the root, and since port allocation isn't fully
dynamic, it's important to run `yarn test` with `lerna run`'s `--concurrency=1`
option. For the same reasons, the root `ci` and `qa` scripts are similarly
restricted. In the future, this setup can be refactored and improved, along
with use of `lerna run`'s `--since` option to increase testing efficiency in
CI.
[PR1318]: https://github.com/embark-framework/embark/pull/1318
Yarn merges `.yarnrc` from the current working directory with `.yarnrc` files
higher in the directory tree, so that file isn't needed in each package.
Unfortunately, npm doesn't do the same for `.npmrc` files, so a similar cleanup
isn't possible.
Add version info to `test_app/extensions/embark-service/package.json` to that
`yarn install` can work correctly if manually invoked in `test_app`.
TL;DR
=====
`yarn install` in a fresh clone of the repo.
`yarn reboot` when switching branches.
When pulling in these changes, there may be untracked files at the root in
all/some of:
```
.embark/
.nyc_output/
coverage/
dist/
embark-ui/
test_apps/
```
They can be safely deleted since those paths are no longer in use at the root.
Many of the scripts in the top-level `package.json` support Lerna's [filter
options]. For example:
`yarn build --scope embark` build only `packages/embark`.
`yarn build --ignore embark-ui` build everything except `packages/embark-ui`.
Scoping scripts will be more useful when there are more packages in the
monorepo and, for example, `yarn start` doesn't need to be invoked for all of
them while working on just a few of them simultaneously, e.g `embark` and
`embarkjs`.
It's also possible to `cd` into a particular package and run its scripts
directly:
```
cd packages/embark && yarn watch
```
Hot Topics & Questions
======================
What should be done about the [README][embark-readme] for `packages/embark`?
Should the top-level README be duplicated in that package?
Lerna is setup to use [Fixed/Locked mode][fixed-locked], and accordingly
`packages/embark-ui` is set to `4.0.0-beta.0`. The same will be true when
adding embarkjs, swarm-api, etc. to the monorepo. Is this acceptable or do we
want to use [Independent mode][independent]?
Scripts
=======
If a package doesn't have a matching script, `lerna run` skips it
automatically. For example, `packages/embark-ui` doesn't have a `typecheck`
script.
`yarn build`
------------
Runs babel, webpack, etc. according to a package's `build` script.
`yarn build:no-ui` is a shortcut for `yarn build --ignore embark-ui`.
`yarn ci`
---------
Runs a series of scripts relevant in a CI context according to a package's `ci`
script. For `packages/embark` that's `lint typecheck build test package`.
Also runs the `ci` script of the embedded `test_dapps` monorepo.
`yarn clean`
------------
Runs rimraf, etc. according to a package's `clean` script.
`yarn globalize`
----------------
Makes the development embark available on the global PATH, either via
symlink (Linux, macOS) or a shim script (Windows).
`yarn lint`
-----------
Runs eslint, etc. according to a package's `lint` script.
`yarn package`
--------------
Invokes `npm pack` according to a package's `package` script.
`yarn qa`
---------
Very similar to `ci`, runs a series of scripts according to a package's `qa`
script. The big difference between `ci` and `qa` is that at the top-level `qa`
first kicks off `reboot:full`.
There is a `preqa` script ([invoked automatically][npm-scripts]), which is a
bit of a wart. It makes sure that `embark reset` can be run successfully in
`packages/embark/templates/*` when the `reboot` script invokes the `reset`
script.
The `qa` script is invoked by `yarn release` before the latter proceeds to
invoke `lerna publish`.
`yarn reboot`
-------------
Invokes the `reset` script and then does `yarn install`.
The `reboot:full` variant invokes `reset:full` and then does `yarn install`.
`yarn release`
--------------
Works in concert with [lerna publish], which will prompt to verify the version
before proceeding. Use `n` to cancel instead of `ctrl-c` as `lerna publish` has
been seen to occasionally misbehave when not exited cleanly (e.g. creating a
tag when it shouldn't have).
```
yarn release [bump] [--options]
```
* `[bump]` see [`publish` positionals][pub-pos] and [`version`
positionals][ver-pos]; an exact version can also be specified.
* `--preid` prerelease identifier, e.g. `beta`; when doing a prerelease bump
will default to whatever identifier is currently in use.
* `--dist-tag` registry distribution tag, defaults to `latest`.
* `--message` commit message format, defaults to `chore(release): %v`.
* `--sign` indicates that the git commit and tag should be signed; not signed
by default.
* `--release-branch` default is `master`; must match the current branch.
* `--git-remote` default is `origin`.
* `--registry` default is `https://registry.npmjs.org/` per the top-level
[`lerna.json`][lerna-json].
To release `4.0.0-beta.1` as `embark@next` (assuming version is currently at
`4.0.0-beta.0`) could do:
```
yarn release prerelease --dist-tag next
```
For *test releases* (there is no longer a `--dry-run` option) [verdaccio] and a
filesystem git remote can be used.
Condensend instructions:
```
mkdir -p ~/temp/clones && cd ~/temp/clones
git clone git@github.com:embark-framework/embark.git
cd ~/repos/embark
git remote add FAKEembark ~/temp/clones/embark
```
in another terminal:
```
npm i -g verdaccio && verdaccio
```
in the first terminal:
```
yarn release --git-remote FAKEembark --registry http://localhost:4873/
```
`yarn reset`
------------
Invokes cleaning and resetting steps according to a package's `reset`
script. The big difference between `clean` and `reset` is that `reset` is
intended to delete a package's `node_modules`.
The `reset:full` variant deletes the monorepo's top-level `node_modules` at the
end. That shouldn't be necessary too often, e.g. in day-to-day work when
switching branches, which is why there is `reboot` / `reset` vs. `reboot:full`
/ `reset:full`.
Errors may be seen related to invocation of `embark reset` if embark is not
built, but `reset` will still complete successfully.
`yarn start`
------------
Runs babel, webpack, tsc, etc. (in parallel, in watch mode) according to a
package's `start` script.
`yarn test`
-----------
Run mocha, etc. according to a package's `test` script.
The `test:full` variant runs a series of scripts: `lint typecheck test
test_dapps`.
`yarn test_dapps`
-----------------
Runs the `test` script of the embedded `test_dapps` monorepo.
The `test_dapps:ci` and `test_dapps:qa` variants run the `ci` and `qa` scripts
of the embedded `test_dapps` monorepo, respectively.
`yarn typecheck`
----------------
Runs tsc, etc. according to a package's `typecheck` script.
Notes
=====
`npx` is used in some of the top-level and package scripts to ensure the
scripts can run even if `node_modules` is missing.
[`"nohoist"`][nohoist] specifies a couple of embark packages because
[`restrictPath`][restrictpath] is interfering with access to modules that are
located in a higher-up `node_modules`.
All dependencies in `packages/embark-ui` have been made `devDependencies` since
its production build is self-contained.
`packages/embark`'s existing CHANGELOG's formatting has been slightly adjusted
to match the formatting that Lerna will use going forward (entries in the log
haven't been modified).
Lerna will generate a CHANGELOG at the top-level and in each package. Since
we're transitioning to a monorepo, things may look a little wonky with respect
to old entries in `packages/embark/CHANGELOG.md` and going forward we need to
consider how scoping our commits corresponds to member-packages of the
monorepo.
In `packages/embark`, `test` invokes `scripts/test`, which starts a child
process wherein `process.env.DAPP_PATH` is a temporary path that has all of
`packages/embark/dist/test` copied into it, so that paths to test
helpers/fixtures don't need to be prefixed with `dist/test/` and so that a
`.embark` directory doesn't get written into `packages/embark`.
The `"engines"` specified in top-level and packages' `package.json` reflect a
node and npm pair that match (a source of confusion in the past). The pair was
chosen according to the first post v5 npm that's bundled with node. A
`"runtime"` key/object has been introduced in `packages/embark/package.json`
which is used as the basis for specifying the minimum version of node that can
be used to run embark, and that's what is checked by `bin/embark`.
Some changes have been introduced, e.g. in `lib/core/config` and
`lib/utils/solidity/remapImports` so that it's *not* implicitly assumed that
`process.env.DAPP_PATH` / `fs.dappPath()` are the same as
`process.cwd()`. There are probably several++ places where that assumption is
still in effect, and we should work to identify and correct them.
`embark reset` now deletes `embarkArtifacts/` within a dapp root, and
`embarkArtifacts/` is git-ignored.
`lib/core/env` adds all `node_modules` relative to `process.env.EMBARK_PATH` to
`NODE_PATH` so that embark's modules can be resolved as expected whether
embark's `node_modules` have been deduped or are installed in npm's flat
"global style".
`checkDependencies` has been inlined (see `lib/utils/checkDependencies`) and
slightly modified to support dependencies that have been hoisted into a
higher-up `node_modules`, e.g. as part of a yarn workspace. eslint has been
disabled for that script to avoid more involved changes to it.
`test_apps` is not in `packages/embark`; rather, there is `test_dapps` at the
top-level of the monorepo. `test_dapps` is an embedded monorepo, and its `ci` /
`qa` scripts `npm install` embark from freshly built tarballs of the packages
in the outer monorepo and then use that installation to run `embark test` in
the dapps. This should allow us to rapidly detect breakage related to
auto-bumps in transitive dependencies.
[filter options]: https://github.com/lerna/lerna/tree/master/core/filter-options
[embark-readme]: https://github.com/embark-framework/embark/blob/build/lerna/packages/embark/README.md
[fixed-locked]: https://github.com/lerna/lerna#fixedlocked-mode-default
[independent]: https://github.com/lerna/lerna#independent-mode
[npm-scripts]: https://docs.npmjs.com/misc/scripts
[lerna publish]: https://github.com/lerna/lerna/tree/master/commands/publish
[pub-pos]: https://github.com/lerna/lerna/tree/master/commands/publish#positionals
[ver-pos]: https://github.com/lerna/lerna/tree/master/commands/version#positionals
[lerna-json]: https://github.com/embark-framework/embark/blob/build/lerna/lerna.json#L11
[verdaccio]: https://www.npmjs.com/package/verdaccio
[nohoist]: https://github.com/embark-framework/embark/blob/build/lerna/package.json#L52-L55
[restrictpath]: https://github.com/embark-framework/embark/blob/build/lerna/packages/embark/src/lib/core/fs.js#L9