Hi, we are using FlyteFiles a lot in our workflows and sometimes we need to access the remote_source property of the FlyteFile. When running workflows locally, remote_source is None, which means we have to explicitly check "if remote_source is None -> use .path instead". To make local execution of flyte workflows be as close as possible to the remote execution, in my opinion it would be a better user experience to have remote_source locally contain the same value as .path. The same strategy is applied when calling .download() i.e. it returns the local path, instead of actually downloading something. I would add this to the list of FlyteFile UX improvements (https://github.com/flyteorg/flyte/issues/4542), but first would like to hear some opinions and if there are some technical implications. Thanks.

That should not be the case? It should have the remote surrender to a local path

I just verified again:

from flytekit import workflow, task
from flytekit.types.file import FlyteFile

@task
def t1() -> FlyteFile:
    filename = "myfile.txt"
    with open(filename, "w") as f:
        f.write("ABC")
    return FlyteFile(filename)

@task
def t2(inp: FlyteFile) -> None:
    print("Remote source: ", inp.remote_source)
    print("Path: ", inp.path)
    print("Download: ", inp.download())


@workflow
def wf():
    out1 = t1()
    t2(inp=out1)


if __name__ == '__main__':
    wf()

Output:
Remote source:  None
Path:  /tmp/flyte-kucg8k7f/raw/88258869af6a3ac5e652f585207a45fe/myfile.txt
Download:  /tmp/flyte-kucg8k7f/raw/88258869af6a3ac5e652f585207a45fe/myfile.txt

Will add it to the list!

cc @broad-monitor-993 is this something you did, where FlyteFile is only set on path?

Not sure, this PR may have caused this behavior: the intent was to use the original file/directory path on local executions instead of copying it over to a temporary location