Thomas Robitaille

Here is an easy and fast way around it: https://github.com/glue-viz/glue/pull/1641/files#diff-4e6dc6cf25762999955424dd2d995973R300 Could we put this kind of logic in bottleneck?

Just wanted to say that I would find this feature very useful for making https://github.com/astropy/reproject more compatible with dask inputs!

For my use case (https://github.com/astropy/reproject), the coordinates might have a similar number of elements as the input array - we are essentially transforming images to different coordinate systems which can...

The hidden error line was: ``` [BinstarError] file /Users/tom/miniconda3/conda-bld/osx-64/glue-core-0.11.0.dev20170705112151.3ea95319-py36_0.tar.bz2 does not exist ``` What's not clear is why this was not shown in the Travis logs (had to run locally...

This is somewhat similar to the issue that when reading a 2D image from a FITS file, the Numpy axis order is reversed from the WCS. I would personally be...

Whatever decision is made should be justified in the specutils documentation - I personally do not understand at all the benefit of having the spectral axis be last. I could...

``swapaxis`` is definitely not the right answer though, as the order will be really whacky for a 4D WCS! (where now it will no longer be simply 'reversed')

Just to add to my comment about 4D WCS, a dataset in Numpy order (spec, time, dec, ra) will now be (ra, time, dec, spec) which seems very confusing.

The issue is that the example suggests the name ``mysql-database`` for the database name in the config step: ``` inveniomanage config set CFG_DATABASE_NAME mysql-database ``` it should be made clearer...

The error disappears if I downgrade flask-collect to 1.2.2