bug in benchmark - but would fix break history? about astropy-benchmarks HOT 5 CLOSED

It's obviously a bug.
These benchmarks were added when rewrite of CompoundModel started. It became clear later that the rewrite was not only performance oriented but also fixed issues related to previous design. In that sense the history before that (4.0) is not interesting, we cannot go back to that version of the code.

I'm fine with either action but 1) is simpler and we don't loose much.

from astropy-benchmarks.

I think option 1 is the easiest, just rename 'large' to 'big'? Then later we still have the option to go back and compute it for older commits if we want.

from astropy-benchmarks.

I think we have 2 choices but @astrofrog should decide:

1. Rename and lose history.
2. Keep same name and @astrofrog re-write history. (If I remember correctly, he said it takes a few days.)

from astropy-benchmarks.

Also looks like @nden added it in #51, so maybe she could clarify.

from astropy-benchmarks.

@hamogu , would you be interested to submit a PR to rename the test + bug fix? 😄

from astropy-benchmarks.