-
Notifications
You must be signed in to change notification settings - Fork 0
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
Prototype migration with generic dataobject #2
base: main
Are you sure you want to change the base?
Prototype migration with generic dataobject #2
Conversation
* runtime already is established and it can communicate back to the runtime factory to tell it to create a migration summarizer. | ||
* | ||
* A thought I'm concerned about is that the search blob can disappear for all dataObjects that are modified, though search will | ||
* resume on the next summarization (probably). |
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.
Since the app implements the DataMigrationObject, perhaps the app can regenerate and include the search blob as part of the migration step?
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.
Yeah, I think they can definitely do that.
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.
I think the thing I worry about is that they have their own implementation of the fluid runtime, it would be nice if they didn't need to do that.
I think it would be better if we provided a better API where they just call into the datastore runtime to pregenerate the search blob
🔗 Found some broken links! 💔 Run a link check locally to find them. See linkcheck output
|
…re, runtime, Loader in versioned tests (microsoft#20650) An extraction from microsoft#20621 Reason for changes These tests explode when I change interface between DDS & DataStore Description These tests are not written correctly. They mix arbitrary version of DDS with the latest version of FluidDataStoreRuntime (through TestFluidObjectFactory). We never supported such mixes - version of DDS has to match version or data store! Using matching TestFluidObjectFactory is easy, but then we start running into somewhat similar problem of mixing FluidDataStoreRuntime & ContainerRuntime. We support only N/N-1 mixes here, and nothing more. So we have to use the version provided by test framework, thus you see some dance above CodeLoader to get right factory. And while we are at it, it would be great for the test to use Loader version provided by test framework, not latest version. This breaks for old versions as the way test is written it works only for latest structure of the document. Previously test framework was running a ton of combos, but most of them were actually the same, as even when it was supplied different versions of loader, we were using only latest version of loader. So, the right fix here - use loader version provided by framework but skip all the tests that use old version of the loader, thus getting same result, but with fewer iterations (and faster tests). As part of fixing it, use proper back-compat compatible ways to get to entry points at loader & runtime layers.
## Description This makes 3 changes to Tree.is: 1. calling Tree.is(node, SchemaNotUsedInTreeSchema) is no longer an error. This undocumented edge case could have been problematic, and is inconsistent since it would not fire if node was not a TreeNode. 2. Tree.is now takes in ImplicitAllowedTypes, making cases like Tree.is(x, [schema.number, schema.string]) valid. THis is more performant and more concise then doing two separate checks and ORing them together. This also allows checking a node against an ImplicitAllowedTypes pulled from a FieldSchema which could be handy for some generic code. 3. The implementation of Tree.schema and Tree.is have been rewritten to not rely on flex-schema as much, and fast path non-TreeNode inputs. This should make it more maintainable and more performant. Interestingly `#2` above is the only case that couldn't be covered by `instanceof` (assuming TypeScript 5.3): we could make `instanceof` do all narrowing currently done with Tree.is, except for this new case. The presence of this case thus seems to motivate keeping `Tree.is` if for no reason other than it can support this additional pattern which `instancof` cannot.
AB#5229