Make the test suite more robust

[hameerabbasi]

I was thinking of moving the tests to a more robust unit test model. It'll take:

• The dtype, shape of each input array.
• The function name to test (both sparse and NumPy/SciPy) or callable.
• Where to place the arguments.
• Any additional arguments to the function.

It will:

• Generate the random arrays.
• Perform the operation on both arrays.
• Compare/assert.

This way, tests just reduce to "stub classes" instead of repetitions.

cc: @mrocklin

[mrocklin]

No strong thoughts from me. Nice testing suites are nice.

…

On Thu, Jun 28, 2018 at 12:14 PM Hameer Abbasi ***@***.***> wrote: I was thinking of moving the tests to a more robust unit test model. It'll take: - The dtype, shape of each input array. - The function name to test (both sparse and NumPy/SciPy) or callable. - Where to place the arguments. - Any additional arguments to the function. It will: - Generate the random arrays. - Perform the operation on both arrays. - Compare/assert. This way, tests just reduce to "stub classes" instead of repetitions. cc: @mrocklin <https://github.com/mrocklin> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#163>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AASszHkzLRduNy4XqdaBQCAiC7vFq94Hks5uBQDbgaJpZM4U7rgH> .

[rgommers]

Sounds like that could be helpful. The one thing to check is designing it in such a way that the tracebacks are still as understandable (exactly which set of arguments is failing)