The module paths are split by ':' and with a Windows filepath (and the 'C:/' prefix),
the split returns the drive letter and directory path separately.
This causes the modulePath to be set incorrectly into the packagePaths object
and the subsequent call for the licenses by key returns undefined on Windows only.
This commit recombines the drive and directory paths and sets the correct key into packagePaths.
* [indexPattern/create] fix test subject
* [indexPattern/create] only auto-select when there is more than one time field
* [indexPatterns/create] describe weird timeField auto-selection
* [indexPattern/create] update no-date-field text
* restore functional test changes in 02787e3df5
(cherry picked from commit 9f0caba9b6)
* Make Discover field chooser items keyboard accessible.
* Make records count link and plus/minus icons tabbable.
* Prevent scrolling when you hit spacebar to toggle a field.
* Add accessibleClickKeys service and kbnAccessibleClick directive, with tests.
* [babel] tweak babel options for performance and debugging
* [babel] remove babel-plugin-transform-export-extensions
* [npm] lock new deps to the patch version
* [npm] remove unused deps
(cherry picked from commit 36ecf130ad)
- Apply aria-label to field name icons.
- Change title to alt for Discover 'Back to top' link.
- Remove unnecessary title attributes from Discover No Results state.
- Add aria-label to Discover sidebar's string field value.
- Adjust aria-label of Discover Field Chooser 'Show field settings' button.
* [scripts] convert `npm run makelogs` to script
* [scripts] convert `npm run mocha` to script
* [scripts] add warning to `test:ui:runner` npm script
* [script] fix location of scripts in warnings
(cherry picked from commit d126e7c5f4)
Test randomization was marking cgroups as unavailable so subsequent tests would fail. In lieu of exposing a reset method, I felt it would be better to refactor out into a class and use an instance method to track when cgroups is unavailable.
Signed-off-by: Tyler Smalley <tyler.smalley@elastic.co>
* Don't kill an entire dashboard because of one bad request
Some tests get rid of the angular promise library so it works better if
it’s a separate function then on the actual angular Promise class.
* Fix promise path references after creation of index file.
* Remove index suffix from import paths
* Remove promiseMapSettled
Use a much simpler implementation
* Clean up
* Hide the second toast when adding a new visualization straight from dashboard
* fix tests that assume two notifications
* fix another tests that assumes two toasts
Since I'm considering not doing a full blown re-write of the existing filter model I wanted to get this taken care in a smaller way so we're not taken by surprise when the deprecated syntax disappears in 6.0.
* using ui-select for field selection in visualize
* adding a limit and infinite scroll
* fixing based on CJ's comments
* Add sort prefix first utility
* Add sortPrefixFirst tests
* updating uiSelect to use the new filter
* rebasing on master
* adding uiSelect to top hits sort-on input
* fixing tests
* [ftr] remove digdug, use chromedriver directly
why?
- digdug is meant to be used by intern, and expects a certain lifecycle that the FTR has tried and continuously fails to mimic
- rather than continue trying to force digdug into the stack, just spawn chromedriver ourselves. We know how to handle it
- cleans up verbose remote logging while we're in there, since selenium-standalone-server went with digdug, which was previously doing the verbose logging
- deprecate config.servers.webdriver
- add config.chromedriver.url
- if url at config.chromedriver.url points to a server that responds to http requests use it as the chromedriver instance, enables running chrome in a VM or container
- if pings to config.chromedriver.url fail a local chromedriver is started
* address review requests
(cherry picked from commit f76bef46c0)
* Create CollapseButton component class to standardize appearance of this button.
* Fix positioning of LocalSearch icon.
* Update collapsible-sidebar directive and Discover page object to use test subject selector.
* Refactor 'closed' class assignment.
* Hide dotted line around focused select component in Firefox.
* Use correct font-family for TextArea component in Firefox.
* Fix gray background of StaticInput in Firefox.
* Reactify the confirmation modal
Up next: jest tests
* Add tests
- Relies on https://github.com/elastic/kibana/pull/10821 getting
checked in first for commonHtmlProps
* Don't include the overlay as part of the confirm modal component
* Use the react version of a confirmation modal
- Can’t use the modalOverlay or it would be two nested react roots, due
to the way it’s embedded in angular.
* Add snapshots
* Fix tests
* fix confirm_modal_promise tests
I have no idea why the introduction of react would cause this to fail
as it was, but for some reason, grabbing the button reference has to be
after the digest loop.
* Address code review comments
* switch over the remaining React.PropType references (in the modals dir anyway)
* Add back dashboard descriptions
Fixes https://github.com/elastic/kibana/issues/10158
* Don't make the description a link
I think it looks weird like that
* wrap long descriptions in the table
* Fix style issue
* Only show one sort icon
* Remove the _source field toggle button which was accidentally reinstated
* Use ng-if instead of css class to hide Remove button on _source field list item
