@apappascs The PR build fails - can you please take a look? https://github.com/spring-projects/spring-ai/actions/runs/15634414758/job/44072094143

sobychacko

thank you @sobychacko I am working on that. need to disable a few tests.

@apappascs Curious why you disabled those tests. Did your changes make those tests fail? In that case, we need to investigate why that is the case and address it accordingly. If we leave these tests disabled, we risk missing regressions that would normally be caught by these tests, including failures unrelated to the current changes.

@apappascs Curious why you disabled those tests. Did your changes make those tests fail? In that case, we need to investigate why that is the case and address it accordingly. If we leave these tests disabled, we risk missing regressions that would normally be caught by these tests, including failures unrelated to the current changes.

Very valid points @sobychacko . I sould have linked this issue
#1193
for clarity. The retry method used for blocking calls isn't working as expected for the streaming calls. The tests are related to the code changes I did and have been disabled or removed for the other models as well.

@chemicL thoughts?

@chemicL thoughts?

#1193 (comment)
I am not sure whether this PR tackles any real problem? The retry template should not have any effects on reactive chains, unless their implementations throw, which should be a programming error. So removing them should not have a significant effect, the bulk of the work is the strategy for retries - whether the framework implements it or it's the user burden.

Thank you for the input @chemicL . I think this alone, as you mentioned, does nt really do much. We discussed last year about removing the retry from streamable calls because it didn't work properly as we expected.

I followed the same pattern we have for the non streamable calls:

#1193 (comment)

And your comment added to the context:
#1193 (comment)

@markpollack @tzolov This is still open.