python-tuf icon indicating copy to clipboard operation
python-tuf copied to clipboard

metadata API: serialization is missing a final newline

Open jku opened this issue 2 years ago • 2 comments

Our JSON serialization is basically json.dumps(metadata_obj.to_dict(), ...).encode("utf-8")

json.dumps() return value does not end in a newline. This means any files written with this content will not end in a newline: this makes the produced files not valid posix text files.

For Metadata.to_file() this seems like a minor bug (and Metadata.to_bytes() should have the same output for consistency), so in practice JSONSerializer.serialize() probably should add the final newline.

My only concern about this is that it is an unexpected change in the serialization output... IF we change this, we should definitely consider this an API change for users who produce metadata.

jku avatar May 16 '23 11:05 jku

Hi @jku, what is the conclusion on this, we want to do this change? If yes, I would like to make it.

h4l0gen avatar May 28 '24 18:05 h4l0gen

I'm worried this will break things downstream (at least all kinds of test suites in applications, maybe even actual application code)... so maybe not worth fixing unless someone is willing to do a lot of analysis/testing in the applications.

jku avatar May 29 '24 06:05 jku