Brian McFee

> Polish up the transform class so that data limits are propagated correctly Following up on this: it turns out to be somewhat complicated. My implementation above implements f0 displacement...

Confirmed that option (2) is not viable, so we may have to scrap the transform idea for now. If we want this feature, it looks like the only way forward...

> Confirmed that option (2) is not viable, so we may have to scrap the transform idea for now. Following up on this, the precise issue is that the `fill_between`...

Still thinking this one over, and I'm ambivalent about how to go with this feature. Should we implement a hacky version of it for the 0.9 release, or punt it...

I think I'd like to punt this feature (again). There doesn't seem to be an urgent need for it, so I'd rather wait a bit and get it right than...

For followup, the matplotlib PR we'd need for this is https://github.com/matplotlib/matplotlib/pull/15595

Looks like the matplotlib PR is off the docket for 3.6, so I'm removing this feature from the 0.10 milestone. :cry:

This issue should be blocked until #745 is resolved.

We had a couple of scipy sprinters looking into this one, but they didn't comment here and I don't have their contact information. (If this is you, please chime in!)...

Started digging into this and it's a bit subtler than I anticipated. I'm pretty sure we can bring the memory footprint down here with a combination of clever fft usage...