In order for the runbld project we use in CI to auto-discover the junit reports we generate they need to match `TEST-*.xml`, which will allow us to include more details in the failure emails, and push failure info to the build-stats cluster.
(cherry picked from commit 8d0b7e7f7e)
Previously, we were not aggressive in combining common code which resulted in duplicates included in the bundles.
As an example `node_modules/elasticsearch-browser/elasticsearch.angular.js` is present in the following chunks:
* kibana.bundle.js
* dashboardViewer.bundle.js
* apm.bundle.js
* monitoring.bundle.js
* ml.bundle.js
* timelion.bundle.js
* graph.bundle.js
Vendored code (anything inside node_modules) is placed in vendors.bundle.js while everything else with more than two references is placed in commons.bundle.js.
This has a couple positive side-effects (numbers are with x-pack & canvas):
* Decreased build time. Seeing builds go from 475.76 seconds to 274.72.
* Decreased memory overhead. Uses roughly 1/3 the memory overhead.
* Decreased bundle size. A 68% reduction in overall bundle size. Going from 66.16 MB to 21.13 MB.
Signed-off-by: Tyler Smalley <tyler.smalley@elastic.co>
* Switch to Yarn (#15485)
* switch to yarn
* cleanup misc references to npm
* [yarn] loosen dependency ranges so yarn will merge more deps
* fix linting error now that moment uses ESM
* [licenses] font-awesome changed the format of its license id
* Use local yarn
* Misc fixes
* eslintignore built yarn file
* Remove mkdir which doesn't do what it should do
* Check build without upgrading lots of versions
* Fix license check
* too many moments
* Better description
* Review fixes
* Lock to angular@1.6.5
* More specific version locks
* Revert "More specific version locks"
This reverts commit 11ef81102e.
* Revert "Lock to angular@1.6.5"
This reverts commit 3ade68c14c.
* rm yarn.lock; yarn
* Forcing a specific version of React, Angular, Moment
* Using vendored version of yarn in ci
* Use --frozen-lockfile
* fixes
* Update lockfile
Previously, we were not aggressive in combining common code which resulted in duplicates included in the bundles.
As an example `node_modules/elasticsearch-browser/elasticsearch.angular.js` is present in the following chunks:
* kibana.bundle.js
* dashboardViewer.bundle.js
* apm.bundle.js
* monitoring.bundle.js
* ml.bundle.js
* timelion.bundle.js
* graph.bundle.js
Vendor code (anything inside Kibana's node_modules) is placed in vendors.bundle.js while everything else with more than two references is placed in commons.bundle.js.
This has a couple positive side-effects (numbers are with x-pack & canvas):
* Decreased build time. Seeing builds go from 475.76 seconds to 274.72.
* Decreased memory overhead. Uses roughly 1/3 the memory overhead.
* Decreased bundle size. A 68% reduction in overall bundle size. Going from 66.16 MB to 21.13 MB.
Signed-off-by: Tyler Smalley <tyler.smalley@elastic.co>
* [mocha] use custom reporter for legible results in jenkins
* [jest] use custom result processor for legible results in jenkins
* [karma] enable junit output on CI
* [mocha/junitReporter] accept rootDirectory as configuration
* [jest/reporter] use reporters option added in jest 20
* [toolingLog] remove black/white specific colors
* [dev/mocha/junit] no reason for junit to be a "reporter"
* typos
* [dev/mocha/junit] use else if
* [karma/junit] use string#replace for explicitness
* [junit] use test file path as "classname"
* [ftr/mocha] no longer a "console" specific reporter
The tests in master are currently failing regularly because our current browser tests are serious memory hogs. Investigation reveals that nearly every test is retaining all of the memory it causes to be allocated. We have made some progress to being able to diagnose the problems, but we expect that problem to take some serious work to fix. We need a short-term solution though, and this is it.
Rather than modify the bundling process, we will shard the top-level test suites by name. For now, we've created 4 shards, but adding new shards is trivial if we need to.
Sharding is accomplished by creating a murmur3 hash of the top level suite names, then bucketing based on the hash output. If a test suite resolves to shard2, but we are running shard1, we simply never pass the function to `mocha.describe()`. Rather than redefine every describe statement, we have shimmed the global `window.describe()` function to accomplish this.
