Data fixture

[alimanfoo]

This PR addresses #138 by adding a data fixture and a test that reads the fixture data and checks against expectations, to support checks that changes to not break the ability to read previously written data.

[jakirkham]

Would it make sense to keep this data in a separate repo? Just thinking of git's generally poor performance with binary data. Also would recommend squashing before merging. Unfortunately git doesn't really do diffs with binary blobs. So squashing should cutdown on the binary content.

[alimanfoo]

I've reduced the size of the fixture down a bit, git seems to handle it ok. Will squash on merge, thanks for the tip.

…

On Monday, April 24, 2017, jakirkham ***@***.***> wrote: Would it make sense to keep this data in a separate repo? Just thinking of git's generally poor performance with binary data. Also would recommend squashing before merging. Unfortunately git doesn't really do diffs with binary blobs. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://github.com/alimanfoo/zarr/pull/146#issuecomment-296841978>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAq8Qh52yGKugYcxPFXB4p3YbkMljbNGks5rzSSHgaJpZM4NGyoN> .

-- Alistair Miles Head of Epidemiological Informatics Centre for Genomics and Global Health <http://cggh.org> The Wellcome Trust Centre for Human Genetics Roosevelt Drive Oxford OX3 7BN United Kingdom Email: alimanfoo@googlemail.com Web: http://purl.org/net/aliman Twitter: https://twitter.com/alimanfoo Tel: +44 (0)1865 287721

[jakirkham]

Sure. Mainly it is cloning the repo fresh were one gets into trouble first. Though getting the data into one blob should help. Also as it seems like the point is not to change the data, I don't imagine this should be something we need to think about too much in the future.

There are some packaging considerations when it comes to data (e.g. Python packaging of it, size considerations, and considerations for any other package artifacts), but some of these can be pushed off until later.

[jakirkham]

Should these go in MANIFEST.in? Not sure if they are required to run the test suite. Thinking about this in the context of the pending release.

[alimanfoo]

Yes these should go in I think, so anyone can run the full test suite after install, although I should check the file sizes.