Pare down python Chat
features
#62
Draft
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR removes some features in
Chat
that predatechatlas
and add a lot of complexity to the implementation without much benefit.For reference, here is the list:
Chat(messages=)
was removed. Usechat.ui(messages=)
instead.Chat(tokenizer=)
was removed. This was only relevant for.messages(token_limits=[])
which was also removed..messages()
. This reflects an overall change approach for maintaining the conversation history sent to the LLM --Chat
should no longer be responsible for maintaining this conversation state, and instead, another stateful object (perhaps the one provided by chatlas, Pydantic AI, LangChain, etc.) should be used to maintain this state. That said,.messages()
can still be useful if you want a reflection of the message state on the client.transform_user_input
andtransform_assistant_response
decorators were removed. This change also reflects the new recommendation of using another stateful object to submit input and retain conversation history..user_input()
'stransform
parameter was removed (as a consequence of the previous point)..set_user_message()
method was officially removed (in favor of.update_user_input()
)NOTE: Although these features are removed here, we should do a
py-shiny
release that deprecates these features first, then removes them full-stop when we move to re-exportshinychat
inshiny
.Follow up
_chat_normalize.py
logic into a more explicit API for transforming provider-specific completion objects into aChatMessage
/ChatMessageDict
. That way.append_message()
et al. can have a strongermessage
type.shinychat
inshiny
, remove the "3rd party" LLM dependencies from shiny'spyproject.toml