GitHub user zyratlo added a comment to the discussion: Texera Agent: issues to prioritize for the release 1.4
I agree with the list, and with @aicam that 3 is the big one. It might be better to split it though, the quick stuff (lower step limit, a cost cap, Stop actually stopping, deleting an agent killing its work) seems doable before the 1.4 cutoff and covers the "user walks away and it burns money" case. The caching work could come after so it doesn't hold that up. For 2, I agree each user should only see their own agents, and not "locked" agents that other people have made. Persistence is definitely needed, but I agree it might not be feasible to finish before the cutoff. For 4, loading the full operator catalog likely makes every request bigger, which kind of works against 3. Maybe the agent could look up operators as it needs them instead if that is possible. Does the newer agent already fix any of these problems (namely 3)? If so it might be easier to bring that over than patch the current one. GitHub link: https://github.com/apache/texera/discussions/9568#discussioncomment-19784172 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
