Dan D'Avella

It's worth noting that [ruamel.yaml](https://pypi.org/project/ruamel.yaml/) appears to handle this case correctly. We have an issue for exploring the use of that package vs `pyyaml` (#391), so this would be a...

@jdavies-st the YAML standard has no concept of what is hashable and what is not. And the standard seems to say that anything that is representable in unicode can be...

Just tested this using the new release of `pyyaml` (version 5.1) and this still appears to be broken.

At the moment, streaming compression is explicitly unsupported. If you try to use compression with a streaming data block, you'll get an error. However, @vmarkovtsev recently added support for LZ4....

This might become easier now with Python 3.7: see #514.

I can't reproduce this in a new conda environment using Py35. Is it possible that you have another version of ASDF installed in the same environment? Maybe in development mode...

A potentially important use case for this would be providing the ability to stream multiple arrays in parallel. The standard allows only a single streamed array in the main file,...

Bottom line is I think we should strive for writing to be an idempotent operation.

It also might make sense to deprecate the `uri` argument to `AsdfFile` (I'm not sure how much this is used anyway) and instead make it an optional argument to `AsdfFile.write_to`....

This is related to #119.