Bohan Qu

> @LastRemote thanks for your effort and patience! To clarify, let's focus on adding TTFT to OpenAIChatGenerator, not OpenAIGenerator, in this PR to keep it simple and aligned with our...

Updated commit and cleaned up git commit history

@vblagoje Yes, it is working and I am using it in mmy applications right now. This does require some code changes in the Langfuse tracer part to track the TTFT...

@vblagoje Here we go: https://github.com/deepset-ai/haystack-core-integrations/pull/1161 Besides the latest merge caused a lint issue, but I am not sure if it is relevant to this PR since it doesn't change any...

Also, here is the complete picture in my use case if anyone is interested: ```python class _ToolService(metaclass=abc.ABCMeta): """ A ToolService refers to a deployed tool (or a selection of tools)...

Hello @sjrl and @anakin87 , sorry for the ping, but may I request some thoughts or updates on this one? I am still trying to work on this and hopefully...

Following up the ContextVars discussions in #1184 : I would prefer to parse the information through meta instead of ContextVars. IMO ContextVars should be better used to track global attributes...

P.S. Parameter and return type definitions in ReST docstrings are currently ignored to avoid things being too complicated.

@anakin87 Sure thing, here is the GH issue: https://github.com/deepset-ai/haystack/issues/9023

> @LastRemote and @julian-risch > > This was, in fact, not so [hard to do](https://github.com/deepset-ai/haystack/compare/main...add_ttft). Forget the recommendation above. We need to simply timestamp the first chunk received from the...