update ulab to 2.1.5 (+ a doc building fix in ulab)

[jepler]

This includes ulab with v923z/micropython-ulab#359 added. We can anticipate that a future ulab release will be made before CP7.0 goes out stable :)

Testing performed: On the unix port, ran ulab.numpy.linspace(0, 10) and the result looked plausible.

The upgraded version of ulab is nearly fully incompatible with 6.x, but is much more compatible with a subset of numpy & scipy. We believe this provides a better path for users of standard Python systems too.

[v923z]

This includes ulab with v923z/micropython-ulab#359 added. We can anticipate that a future ulab release will be made before CP7.0 goes out stable :)

Version 3.0 should not change how ulab behaves, it should only add a new module: v923z/micropython-ulab#327
However, I thought that circuitpython would stay with the legacy branch for the 6.x series. Has anything changed?

[dhalbert]

However, I thought that circuitpython would stay with the legacy branch for the 6.x series. Has anything changed?

This has not changed. The main branch for circuitpython is now the start of 7.0.0. 6.2.0-rc.0 was released this week, and we created a new 6.2.x branch for it. We hope 6.2.0 final is imminent.

[dhalbert]

I have been looking forward to this - thanks!

My change is suggested, your choice. I'll approve if you don't want to do it now.

[v923z]

However, I thought that circuitpython would stay with the legacy branch for the 6.x series. Has anything changed?

This has not changed. The main branch for circuitpython is now the start of 7.0.0. 6.2.0-rc.0 was released this week, and we created a new 6.2.x branch for it. We hope 6.2.0 final is imminent.

@dhalbert Thanks for the clarification! By the way, if you think that v923z/micropython-ulab#327 has any relevance for circuitpython, you could comment there. This is the lazy-loading module that I mentioned during one of the weekly meetings, and we can still change the interface. I would actually appreciate any input.

[dhalbert]

👍