I think this makes more sense then running them via a separate command,
since that command would only make sense for atom-core, unlike the
current 'run-package-specs' command, which works for any atom package.
This way, they won't run by default, but you can opt in to running them
on the command line by setting an env var, or in the spec runner by
temporarily editing the code, like we do for focused tests anyway.
Previously an editor that was split into multiple panes would not
prompt to save correctly when the window was unloading.
This adds a new `windowCloseRequested` option passed through from the
beforeunload handler to the editor so that it can specially handle this
case.
Closes#5257
This commit also adds to the list of required measurements and updates
the spec with a buildPresenter helper to more easily supply default
values for required measurements in each spec when they aren’t relevant
to that spec’s content.
The release notes are provided by Squirrel on Mac but not by Squirrel for
Windows and the release notes package pulls them down manually anyway
so this field is no longer needed.
Checking for the presence of the release notes previously was preventing the
event from firing on Windows which would cause old release notes to show in
the package.
Closes#3757
This test performs random operations on the editor and assigns random
measurements from the view. After each operation, the state of a
pre-existing presenter is compared with that of a new presenter created
with the same parameters.
Since it’s easier to reason about building fresh state than it is to
reason about state updates, I hope this will catch any bugs in our
update logic as we optimize it and explore every corner case.
This means we have some duplicated values in different parts of the
tree, but it’s cleaner in the view since each component only consumes
a single object. Seems like the presenter should convey the correct
data to the correct locations and minimize the logic in the view. A
few duplicated integers is a reasonable trade-off.
