New RunLLM response generated: >>>*# Question* Ah i found finally the correct location to look at: Here the docstring says, that only other python files are copied, from that flyte tasks are imported. since data.py had a data processing task, but of course the types.py file did not, that one seems to be ignored? I am not clear on why not all imported files are copied over? In which case can other files be imported that are not needed at remote side? They will always produce an import error, so it seems a bit strange, this selection? # Answer Flyte's current behavior is to only copy over Python files that contain Flyte tasks or workflows when using

pyflyte run --remote

. This selective copying is intended to minimize the amount of data transferred and to ensure that only necessary code is included in the remote execution environment. However, this can lead to issues if other required modules (like

types.py

in your case) are not copied, resulting in import errors. According to the Flyte GitHub issue #5492, the current workaround is to use the

--copy-all

flag to ensure all local files are copied over. This approach, while not ideal, ensures that all dependencies are available in the remote environment. Would you like more details on how to use the

--copy-all

flag or any other aspect of Flyte's file handling? Sources: - Flyte GitHub issue #5492 - Flyte GitHub issue #5343 Slack Link https://flyte-org.slack.com/archives/CP2HDHKE1/p1723030922.571689 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.