Revamp the runtime caching option conversion. #1160
Conversation
|
Changes Unknown when pulling 7e21126 on runtime-caching-options-changes into ** on v3**. |
| }), | ||
| broadcastCacheUpdate: joi.object().keys({ | ||
| channelName: joi.string().required(), | ||
| options: joi.object(), |
There was a problem hiding this comment.
I believe the broadcast cache update object needs a channel name.
There was a problem hiding this comment.
Yup—that requirement is enforced one line up, on line 46.
| maxAgeSeconds: joi.number().min(1), | ||
| }).or('maxEntries', 'maxAgeSeconds'), | ||
| networkTimeoutSeconds: joi.number().min(1), | ||
| plugins: joi.array().items(joi.object()), |
There was a problem hiding this comment.
How does this work? I.e. how does code in the node space get into the service worker? My main consideration here is that if this works, why support one way in the build process and a different way in the service worker?
There was a problem hiding this comment.
The code that converts whatever is configured in the node build tools into a string representation that could be inserted into a service worker file is
and the string is used in the template at
(None of that has changed recently.)
The syntax used to generate the service worker code is not exactly the same as the syntax used when executing the service worker runtime. Service worker execution implies that we're inside of the ServiceWorkerGlobalScope with the various definitions in the workbox namespace available. When the build configuration is used, we're inside of the node runtime environment, and we maintain a mapping of the keys in the config that correspond to the resulting runtime code.
I think in this PR the mapping between configuration supplied to node and resulting Workbox v3 service worker runtime code gets much closer to being 1:1, as we removed some differences that were mainly around to match sw-precache's config.
|
Based on the approval, I'm going to merge. Matt, if you've got any larger concerns about the way the |
R: @philipwalton @addyosmani @gauntface
Fixes #1096, which also contains an explanation of what's changing and why.