Closed adrn closed 5 years ago
Note this also addresses astropy/astropy#7521
There is one minor comment, but it shouldn't be the reason for the travis failure, I'm a bit puzzled by it but maybe the benchmark gurus have a good idea about how to fix it.
Ah, yes, that was a bug! But I don't know what could be causing the build failures. @astrofrog ?
Just to clarify my previous comment - if you really want to modify a benchmark, that's of course fine, but you should rename it at the same time to make sure we don't have to go back and remove old results.
Using asv dev --bench="coordinates.*"
I discovered some bugs in the benchmarks (note, @adrn, that seems to be the easiest way to debug the benchmarks themselves). Note sure if that's the same as the CI failure though, so I pushed a commit up really just to see what happens...
I didn't address @astrofrog's suggestion of not modifying, though - will leave that to @adrn.
The benchmarks run for me locally with asv dev
, so I'm not sure why travis is failing. I think I addressed @astrofrog 's comments by backing out the random seed changes. But I did rename some benchmarks for clarity.
The failure is in master and is unrelated
This tests different paths through the SkyCoord initializer in a bit more of a controlled manner.