-
Notifications
You must be signed in to change notification settings - Fork 3.6k
Playground Editor V2: Multi-file, ES modules, NPM Support + Build output hashing #17216
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: master
Are you sure you want to change the base?
Conversation
…styling, inline definitions and navigation to symbols
…nnable compiled by monaco ts worker
I think there may have been something related to during-development serializing of that project right there... I see the issue on that saved Playground, but when I start from scratch and save a new project and try the repro steps it seems to work every time. If you have the chance could you give that a shot? I will take a closer look into exactly what that case it doing there as well. |
It still doesn't work for me. Try with https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/PLAYGROUND/refs/pull/17216/merge/?snapshot=refs/pull/17216/merge#TX2DRJ#2 (I've just created the PG) |
Ha, I just noticed your test module doesn't have any extension, it's just called |
That's totally fine with me, validating the PR! |
Looks like there are some conflicts, I will merge upstream and have changes pushed tomorrow morning. |
Reviewer - this PR has made changes to one or more package.json files. |
You have changed file(s) that made possible changes to the sandbox. https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/SANDBOX/refs/pull/17216/merge/ |
You have made possible changes to the playground. https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/PLAYGROUND/refs/pull/17216/merge/ The snapshot playground with the CDN snapshot (only when available): Note that neither Babylon scenes nor textures are uploaded to the snapshot directory, so some playgrounds won't work correctly. |
Devhost visualization test reporter: |
WebGL2 visualization test reporter: |
Visualization tests for WebGPU |
@ryantrem I merged in your most recent changes around inspectorv2 and validated state hydration for the cases I'm familiar with, would probably need a pass from your side to make sure it's all kosher |
Thanks for the heads up! I tested it and its not working correctly with older versions of Babylon. I think there was a bad merge and the changes from these two PRs did not get merged correctly into your branch:
You can test by using older versions of Babylon by doing the following:
For older versions, there should not be unhandled errors and if you try to enable Inspector v2 you should get an alert saying Inspector v2 can only be used with the latest Babylon. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Inspector v2 with old versions regressed. More info in this comment: #17216 (comment)
Nice, thanks and sorry I missed that merge, was a lot of deviating points. I've got those changes in and tested against current PG and seems to function as intended, let me know if there's anything I missed |
Reviewer - this PR has made changes to one or more package.json files. |
You have made possible changes to the playground. https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/PLAYGROUND/refs/pull/17216/merge/ The snapshot playground with the CDN snapshot (only when available): Note that neither Babylon scenes nor textures are uploaded to the snapshot directory, so some playgrounds won't work correctly. |
You have changed file(s) that made possible changes to the sandbox. https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/SANDBOX/refs/pull/17216/merge/ |
Devhost visualization test reporter: |
WebGL2 visualization test reporter: |
Visualization tests for WebGPU |
I did a quick test of Playground, but it looks like Inspector v2 is not working at all. It looks like there are still merge issues on these lines:
|
Follow-up to the original PR #17175
The different issues pointed out by different people on the BJS forum post point to a caching issue - some scripts were loaded fresh while others were stale and cached by the browser. This led to undefined behavior all around and the page would be effectively "broken" depending on which assets were cached or not, i.e. anything could go wrong and wouldn't necessarily present as a predictable bug.
Previously there were duplicate names from the dist output and monaco worker output which is cached locally in a browser:

index.html does not have a Cache-Control response header and respects the client's request of
max-age=0
- this is good. index.html should always fetch the latest.It also appears there is a CDN purge script running on the Playground deployment script, although I don't have access to the CI yaml. https://dev.azure.com/babylonjs/ContinousIntegration/_build/results?buildId=44262&view=logs&j=275f1d19-1bd8-5591-b06b-07d489ea915a&t=4d2ea636-6e6e-5831-8b5a-8ce5e9ce8bbc
If that's the case for the last operation in that script, good. CDN purge handles the server cache.
Takeaway from the existing deployment pipeline and CDN architecture, nothing needs to change. If we can guarantee unique output from a build each time, there is no room for cache confusion from the client's browser. This only applies to the Playground and I haven't looked into other deployed apps. Cache-Control of 2 days is fine. Eventually this issue would've sorted itself out in the field but 2 days is a long time to wait with undefined/broken behavior.
The additional changes in this PR are consolidated in these two commits:
1e9c765
knervous@bd197e8
These changes ensure hashed output from every build and deploy for relevant bundled js files. The generated entrypoint for
babylon.playground.[hash].js
is injected into the index.html with the WebpackHTMLPlugin. The helper function for fetching scripts does append a query parameter to bust cache, but dependent chunks of scripts do not automatically follow the same convention. Ideally we can get rid of that?t=[timestamp]
and let caching do its business.For anyone testing changes, please check https://playgroundv2.vercel.app which should be the exact hosted bundle set from the deployment build.
For sanity's sake here are three browsers on mobile tested with the current code: Edge, Firefox and Chrome
With editor open: