This makes the required changes needed to make the editor components
and sagas lazy-load to improve the time before the first render of the
app.
This is done by making sure we only import types from the monaco editor
module anywhere other than the saga and the component modules and using
React.lazy to lazy-load the component and start the sagas.
This cuts the main bundle time to less than 1/2 so load time should be
over twice as fast now.
This turns off the default word-based suggestions. This was useful when
we didn't have proper code completion but now can be distracting.
Fixes: https://github.com/pybricks/support/issues/757
This allows importing user-created files and getting code completion
for those imports. This is done by mirroring the dexie-based file
system used by the editor to the emscripten based file system used
by pyodide. In the future, ideally we would have some sort of shared
file system, but this works for now.
Note: completing `from ` doesn't list user files/modules because of
filters in the pybricks-jedi python package but completing
`from my_file import ` does work as expected.
This changes things up a bit so that files are handled by uuid instead
of path. The allows for files to be renamed without breaking things.
Also some improvements with persistence of the view state for each
file is made.
Since we are using the react-i18n babel loader plugin, useI18n can only
be used once per file. Also we have to remember to add the istanbul
ignore next comment each time we use it. By moving this to a common
file, it can be easily copied and pasted when a new submodule is
created and we don't have to remember the rules when calling it.
This starts moving alerts to the same subsystem where they are relevant
instead of putting everything in notifications.
So far, only the explorer file in use error is handled like this.
it is no longer feasible to build the blueprintjs css ourselves, so we
just use all of the default variable values and override things using
selectors instead.
This breaks down reading and writing into low-level and high level sagas.
Opening now takes a lock on the file so that it can only have one writer
or have multiple readers. The low-level read and write sagas use the
file descriptor to determine if the file is still open before actually
performing the action.
Since we are using the web locks api, these locks should work across
browser tabs/windows.
The high-level read/write sagas perform the low-level open, read/write,
close operations in a single call.
The high-level delete and rename sagas are also updated to take locks
on the files while performing the respective operations and will fail
if the file is already being used somewhere else.
This changes how files work. There is now an open function that
translates a path to a file handle (id). Then this handle is used
to perform other actions on the file. The file contents are moved
to a separate table so that the actual file storage is independent
of the database (e.g. in the future, we may use File Access API).
We store a hash of the file contents in the metadata file so we can
detect file changes without having to compare file contents.
We also separate the change and add actions.
David Lechner