Josh Izaac

Ah I see, yes definitely @charmerDark!

I don't believe so (unless I am mistaken) -- it looks like the plugin is still simply building a new numeric program every time: https://github.com/PennyLaneAI/pennylane-qiskit/blob/1e152c7ecea78d2b51267d6a9f75393951449dcc/pennylane_qiskit/qiskit_device.py#L248

@antalszava I'm curious if the caching in the device is superceded by the caching in the `qml.execute` function? This new caching capability is accessible via the new QNode: ```python dev...

Good idea!

Good idea, there is still a bit of leeway though with regards to the resource name. Should the name of the resource be required to match the title of the...

I guess I only really noticed with the 'Documentation' resources. This causes them to be at the top of the list, but doesn't match the HTML title of the linked...

Hey @sduquemesa, it would still be useful long term to port to using PEP517/518, since this is the future of Python packaging, _but_ as you note the reasons for doing...

> Instead of registering new measurements, it would be great to have the possibility to register custom functions as outputs of qnodes. That custom function would internally call a valid...

@dwierichs could you explain a bit further? It could just be the early morning fogging my brain, but I'm not sure I follow!

This is actually how PL used to work! parameter-shift was done at the qfunc level, not the tape level 🙂 The reason we changed was that it ended up being...