This is a major update to `webpack-dashboard` with breaking changes. And a lot of speed and coolness.
## Breaking changes
- Requires node6+.
- Use the webpack internal stats object instead of real bundles. This is faster and with less futzing / parsing on the dashboard's end. Should permanently fix the `no code sections` found errors.
- We've removed command flags / plugin options that no longer apply like `--root`, etc.
### Modules
- Removed `min` and `min+gz` estimated file sizes from modules outputs. This dramatically speeds up the dashboard and reduces CPU usage.
### Versions
- Version skews should be detected more accurately and much faster thanks to dramatically improved heuristics from `inspectpack`.
### Duplicates
- Duplicates now reports on duplicate files that are (1) completely identical, and (2) the same package name and file path but with different sources. The first are low hanging fruit for optimizations in your build -- you've literally got the same thing multiple times. The latter needs to be coalesced at the package dependency level -- something the versions output can help with.
## Issues
- Upgrade hooks to work in webpack1-4 with abstraction. Fixes#240
## Refactoring
- Switch from `chalk` to using Blessed's built-in tags for coloring. This is definitely the way to go as things like table spacing don't work with chalk strings and work like a charm with Blessed tags.
- Update dependencies and remove lots of unused things (like `lodash`, `chalk`, `bluebird` and a whole lot of `babel*` that apparently wasn't used anyways).
- New `inspectpack` is way slimmer and faster, with no more binary dependencies!
- Switch to yarn-only workflow in dev.
- Bump Travis node versions.
Adds `minified` and `gzip` configuration options that can optionally disable the relatively expensive per-module calculations of each by `inspectpack` under the hood.
- Fixes#227
- Adds `minified` and `gzip` plugin options, both defaulted to `true` so we don't need a semver major bump.
- Adds more README documentation and explanations about CPU intensive workloads and the `inspectpack` cache.
Before, it was slightly confusing, because I was quickly scanning the
percentages and didn't know why a module down the list was at 90%,
before realizing it was 90% of its dependent module. Now it should be
more immediately obvious.
