* Add getSelected/setSelected to core/common.js
* Migrate core to use Blockly.common.getSelected/setSelected instead of Blockly.selected
* Migrate demos to use Blockly.common.getSelected/setSelected instead of Blockly.selected
* Remove Blockly.selected
* Use variable instead of calling common.getSelected() multiple times
* Add accessor for selected on Blockly
* Add renaming record for Blockly.selected -> Blockly.common.getSelected/setSelected
* Use Object.defineProperties instead of .defineProperty when
installing get and set accessors for previously-mutable exports.
* Add entries to renamings.js where feasible (i.e., for static rather
than instance properties).
(Don't do blockly.js yet, to avoid merge conflicts.)
* Use goog.module in mocha tests
* Fix compiler warnings
* Make test helpers a module
* Name test modules Blockly.test.*
This is to be more consistent with how non-test modules are named.
Also remove top-level goog.require of TestHelpers (now
Blockly.test.helpers) since requiring a side-effect-less module does
nothing.
* Convert block_test.js and comment_test.js to goog.module syntax
* Address PR comments
* Goog modulify tests
* Goog modulify toolbox helpers
* Fixes imports and moves common tests from workspace_test.js to a helper file.
* Update test deps after rebase
Co-authored-by: Christopher Allen <cpcallen+git@google.com>
This is part of #5153 but is being prioritised because we want remove
the declareLegacyNamespace calls from the core/utils/*.js modules and
then reexport them explicitly via utils.js, and it turns out that
doing so results in the exports object of this module being passed to
Object.freeze - which fails on the global object, which can't be made
non-extensible!
The new name chosen for the former default export is globalThis, since
it is intended to have the same value as the global variable of that
name; see:
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/globalThis
Create a file scripts/migration/renamings.js to collect information about renamed API entries. Start by filling it with the renamings already done to blockly.js.
* Use goog.requireType when importing I* interfaces
Interfaces have no code, so should never be referred to outside of
(JSDoc) comments, and so the modules that define only interfaces never
need to be goog.require'd - goog.requireType is always sufficient.
This commit fixes imports of all modules whose name matches
/(.*\.)?I[A-Z]*/ - i.e., the hungarian-notation named ones in
core/interfaces/.
* Use goog.requireType when only using import for type specifications
Where a module is imported only to used in JSDoc comments it can
(and should) be goog.requireType'd instead of goog.require'd.
* Remove spurious eslint-disable no-unused-vars
There were a few cases where modules were being imported with
goog.require (because they are referred to in code, not just JSDoc
comments) but were prefaced by a spurious eslint suppress.
Remove these, restoring the invariant that an import gets an eslint
if and only if it is a requireType.
* Remove obsolete Closure Compiler error group
stricterMissingRequire has been superceded by missingRequire, and now
causes a Java null pointer exception if supplied.
An intermediate version of this code read:
fs.mkdir(demoStaticTmpDir, {recursive: true}), done);
but apparently `fs.mkdir` doesn't honour the recursive option, so I
tried to revert the change but munged it instead.
This commit cleans up the mess I made.
* New gulp task buildDeps (npm run build:deps) to create tests/deps.js.
* Old gulp task buildUncompressed is deleted.
* blockly_uncompressed.js is now handwritten.
* And simplified; in particular, BLOCKLY_BOOT is gone.
* And linted!
* For consistency with the Closure documentation and base.js,
consistently refer to what blockly_uncompressed.js is used for as
"uncompiled mode" rather than "uncompressed mode" (but don't yet
rename it to blockly_uncompiled.js)
I had forgotten that I needed to change the value of BUILD_DIR
in several different places.
Added comments warning future editors about this as well as filing
issue #5007 to track fixing this properly.
Despite being misconfigured and therefore failing, the typescript
and metadata test scripts were exiting with status 0, indicating
successful completion. These have been fixed so they should fail
on any error, including misconfiguration.
