Gabriel Selzer

This may be of help: https://mail.python.org/pipermail/python-dev/2008-January/076194.html

Hmm, fails on my machine too @elevans. Looks like the object `sj.to_java(narr)` is a `DefaultDataset` with `NativeBoolType` elements, while `rai` is an `ArrayImg`. Naturally, the matching copy Op doesn't handle...

> One small question is: instead of using only `setVisible(true)`, should we also be invoking `requestFocus()` on the already-shown application frame? Oh, and as far as this goes, I think...

I actually would also prefer to do it reactively, like I said in the issue I just wanted to remember to delete that conditional when we eventually **do** need to...

This may be of use: https://doc.qt.io/qt-6/qtreewidgetitem.html#setIcon

I started working on this, but I quickly realized that adding icons to a `QTreeWidget` would be tricky, and figured that switching our search results to a Qt model/view approach...

Yeah, that's tricky, as we are definitely bounded by magicgui/qt here - that's the default min/max for integer types, iirc. We could probably start with a function like [this one](https://github.com/imagej/napari-imagej/blob/75cedb62a87a8d8ec20a1669133c517bf48c0b6e/src/napari_imagej/widgets/parameter_widgets.py#L44)...

If it's an output, then we are creating that widget by ourselves ([here](https://github.com/imagej/napari-imagej/blob/75cedb62a87a8d8ec20a1669133c517bf48c0b6e/src/napari_imagej/widgets/napari_imagej.py#L148)) - we can simply update the widget that is returned.

> I can't remember why we decided that original ImageJ commands should fail when you try to run them, instead telling you that you must run them from the ImageJ...

> For example, napari-imagej could perform a proactive layer sync when the user launches an original ImageJ command (or maybe any command or op? what's the downside?) Note that this...