You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
"message": "Migrate VertexAI serialization to be localized (#6631)\n\nThere are some considerations to how this should be finalized. Current\r\nimplementation details that I've decided on which we can change:\r\n* Based on the example doc, all classes `Foo` have been renamed\r\n`InternalFoo`\r\n* All internal serialization classes, where relevant, are moved to inner\r\nclasses of their API counterparts\r\n* All classes only used as child fields for serialization classes have\r\nbeen moved to inner classes of those classes\r\n* All `toPublic` and `toInternal` methods on API and serialization\r\nclasses have been moved inside of those classes and `conversions.kt` has\r\nbeen mostly emptied.\r\n* A few serialization classes do not have API equivalents and are left\r\nin a `Types.kt` file\r\n\r\nPossible changes:\r\n* Change all `InternalFoo` classes to have the same name, referenced as\r\n`Foo.Internal` rather than `Foo.InternalFoo`. This will probably make\r\nthe codebase feel cleaner, but I'll wait for opinions on it\r\n* Move serialization only classes out of the internal serialization\r\nclasses, either to inner classes of the API classes or top level classes\r\nthemselves.\r\n* For classes that have serializers, rename serializers from\r\n`InternalFooSerializer` to `Serializer` for example\r\n`Foo.InternalFoo.Serializer` or `Foo.Internal.Serializer` instead of\r\n`Foo.InternalFoo.InternalFooSerializer` or\r\n`Foo.Internal.InternalFooSerializer`",
"message": "Update the vertex golden file version being pulled (#6651)\n\nWe've updated from version 3.* to 5.*. Test were added were necessary to\nmatch the coverage in the iOS repo.\n\nA few changes (v5.1 and v5.2) don't have tests covering the new files\nyet",
"message": "Drop empty text parts when parsing the model's response (#6652)\n\nEmpty text parts go from a nuance when processed, to an exception when\nsend to the backend. To prevent this issue, we are dropping them when\nparsing the response from the server.",
0 commit comments