David Frizelle

Similar error occurs at ```io.r2dbc.h2.codecs.BlobToByteBufferCodec.doEncode(BlobToByteBufferCodec.java:65)``` when using Spring Security Oauth2 client configured with H2 and ```R2dbcReactiveOAuth2AuthorizedClientService``` ``` java.lang.IllegalStateException: Iterating over a toIterable() / toStream() is blocking, which is not supported...

This issue disappeared from local app and mui.com, possibly was a browser, rather than mui issue.

An observation - OpenSAML 4.3.1 is on the JBoss Public repo and not on Shibboleth https://mvnrepository.com/artifact/org.opensaml/opensaml-core/4.3.1 https://mvnrepository.com/artifact/org.opensaml/opensaml-core?repo=shibboleth-releases

@tox-p GPT 4 Turbo is production ready. Created PR https://github.com/knuddelsgmbh/jtokkit/pull/94 with the new model name.

Passing a `Filter.Expression` to the `SearchRequest` covers this using metadata. For example: ```java // Adding a document with metadata fields Document doc = Document.builder() .withContent("Sample content") .withMetadata("category", "technology") .build(); vectorStore.add(List.of(doc));...

Left o1-preview out of enum for now as previously jtokkit waits until models get their final name. The `registry.getEncodingForModel` method will work correctly with this PR for both of them...

Confirmed bug and that your suggested fix works. Raised https://github.com/spring-projects/spring-ai/pull/1423 with fix.

Fixed in https://github.com/spring-projects/spring-ai/issues/1537