Open mkhl opened 6 years ago
I added a fix that passes all tests but I’m not certain I’m happy with it yet.
Correction: It doesn’t pass the tests, but the lingering eslint errors hide the failures.
Thank you. I've attempted to build on your efforts in complate:async (note that this expands the corresponding non-renderer test in order to reproduce the issue at a lower level) - mostly hacking aimlessly in post-holiday stupor, which did yield some success though: The async tests are green now, but we're back to recursion errors (cf. cc106c1aa14d2b5119ab86d891f3451744e968e6), which is what I was afraid of with this approach. 😞
FYI: I've now cherry-picked your performance tweak into master (53a936520cc47f91bdf19776fb1e82ec9a2a90dd) and updated complate:async accordingly. Still pondering what all this means though... 🤕
Note: This is a bug report with a test case attached.
We don't currently seem to have a test case that ensures that basic rendering of async elements (in async mode) works correctly, and the attached test fails on master, and not just since cc106c1a but at least back at v0.12.0. (Earlier versions had too different API for me to test easily.)
I may be misunderstanding the macro, but I expect it to render
when instead it renders (note the different nesting):
Note the incorrect nesting: I believe this is due to sibling elements being rendered before a child with asynchronous children has had a chance to finish rendering.