Stuart Berg

Follow-up question: If I only need support for `precomputed` and not for `n5` or `zarr`, is there a way to disable `n5` support to simplify the build? (Currently, the segfault...

Thanks for the tips. I'm still running into issues, but I don't expect you to resolve them for me right now. I can put this aside for a while. For...

For my needs, the error-check you've already proposed is good enough. (Thanks, BTW.) I don't quite follow the explanation of why switching to byte-based strides is more powerful than `sizeof(T)`-based...

As I mentioned in https://github.com/conda-forge/vigra-feedstock/pull/43#issuecomment-436447997, I have concerns about the proposed workaround. I think we should avoid `-undefined dynamic_lookup` if possible. I've had only negative experiences with that linker setting...

+1 for this old issue, at least for TIFF and JPEG. When reading tiles from a remote server, it is nice to open in-memory image files without writing them to...

I think I disagree. I would expect 200 in both cases. Either way, I wouldn’t worry about current apps. I don’t think we depend on 404 behavior. And I don’t...

Indeed, this seems to be fixed in newer builds. (The v0.9.17 build can launch without errors on my Mac.)

Might be nice to have `synweight`, too (`upstream + downstream`).

Here's a [pure-python function](https://github.com/janelia-flyem/neuclease/blob/b29b80eef624d42140f93342190b5a8185fd9b1c/neuclease/dvid/annotation.py#L168) that achieves the same thing, but with the full synapse list from the `synapses` instance.

This seems to be an issue with `labelmap` instances, too. For instance, I can't edit the `MaxDownresLevel` after the instance has been created.