The tests run using TinyWorker, which emulates web workers on Node.
There are quite a few quirks in that situation due to the differences
between Node and Webpack as well as the differences between TinyWorker
and real Web Workers.
The tests also exposed a few bugs in the dispatch system, which have now
been fixed. Most notably, if a method called through the dispatch system
throws an exception that exception will now be passed back to the
caller. Previously the exception would escape the dispatch system and
the caller would never hear any response at all.
Rather than assuming that the storage instance will be attached to a VM,
just return it. Callers may attach it to a `VM` or (in the case of
`import_sb2.js`) to a `Runtime`.
This new test verifies that clones and their associated threads are
cleaned up properly by the `delete this clone` block. The clones run two
stacks each: one which waits and then deletes the clone, and another
which includes a `forever` loop: this is to verify that the thread
running the `forever` loop is ended when the clone itself is deleted.
The project does this with two batches of clones to ensure there are no
problems with reusing array indices previously occupied by now-removed
threads or clones.
When importing a project we know the file extension for each asset to be
loaded. This change provides that information to the storage system so
that we can load assets which don't use the default. For example, this
allows loading JPG-format backdrops.
In support of this change, there's a new function on `StringUtil` called
`splitFirst`, which splits a string on the first instance of a separator
character. This change includes unit tests for this new function.
When no Device Manager instance is present this test takes a long time
and can cause timeouts. We'll need to fake the HTTP requests if we want
to do this sort of test reliably.
* Add costume dataURI to costumes from storage
Towards #515, LLK/scratch-gui#279
* Fix tests
* Load costumes incrementally
Some of our tests assume that at least some of our costume data is available before the costume data is loaded. So, provide as much costume data as is available.
* Remove unnecessary filter for null costumes
Resolved when we updated to load costumes incrementally.
/ht @cwillisf
- Attach the storage module to the VM for tests which load projects or
other assets.
- Move `import_sb2.js` from `test/unit/ into `test/integration` since it
now depends heavily on `scratch-storage`.
- Skip loading costumes when there is no renderer.
* Revert "Merge pull request #486 from rschamp/fix-filter-tests"
This reverts commit ba00db897f, reversing
changes made to 739c5deb63.
* Revert "Show Categories that use custom code to load (variables, procedures) (#483)"
This reverts commit 739c5deb63.
* Revert "Merge pull request #461 from rschamp/filter-toolbox"
This reverts commit 343b5bfe8e, reversing
changes made to 370f2c6a47.
Changed mathop to call new math util
Changed sin & cos to round correctly (to get 0)
Added testing for the new math util function
Added testing for the new mathop functions
Had to update the test scripts to handle the alpha channel, also I note
that all the hex tests are using CSS notation, not scratch notation
(which is 0x not #)
Changes include:
- Added missing JSDoc for items in `scratch3_pen.js` and `target.js`.
- Moved constants used by `Scratch3PenBlocks` into the class.
- Created a constant for minimum and maximum pen size.
- Added `util/clone.js` to host cloning functionality.
- Pen blocks now check for the renderer before trying to use it.
- The pen integration test covers all blocks, though `clear`, `stamp`,
and `pen down` will skip some of their functionality when there is no
renderer.
Also, fix a math error in `Color.rgbToHsv`.
Newly covered functions:
- `toRbgColorObject` from `util/cast.js`
- `hsvToRgb` from `util/colors.js`
- `rgbToHsv` from `util/colors.js`
- `mixRgb` from `util/colors.js`
