[BUG] figure.circle produces empty plot when used datetime axis about bokeh

[bryevdv]

This capability was added in 3.4 in PR #13554

In fact, your problem is a usage issue. As stated in the documentation, the scale of datetime axes is "milliseconds since epoch" In your code you are asking for circles with a 1 millisecond radius on a scale of months. So, entirely predictably, they are "invisible". If you change the radius to be a reasonable number of milliseconds for axis that spans months:

radii = np.ones(N) * 100000000

Then the circles show up, and the dilation as well:

However, I might suggest using scatter rather than circle which would afford configuring a size in pixels, rather that a radius in data units.

Lastly: I am trying to be patient since you seem to be both new to Bokeh and having struggles. But given that, I'd ask you to consider that if you are hitting an issue, it is entirely possible that it is a usage issue. The GitHub issue tracker is reserved for bug reports and feature requests only. It is not the appropriate place to seek general support or help with usage questions. For that, use the Community Discourse site.

[bryevdv]

Also, re:

have you considered that those OSS contributors left because the code base is not friendly to contributors?

You misunderstand, completely. That statement was about me, seriously considering leaving OSS development after nearly 20 years, because I don't want to put up with users who are "pricks by nature" (your words) any longer.

[u3Izx9ql7vW4]

Since you closed my discussion point, have you considered that those OSS contributors left because the code base is not friendly to contributors? I spent the better part of the weekend going through the code base and undisciplined is how I would describe certain aspects of the code base, granted it is OSS but still. To be fair I think Bokeh is the best graphing library out there for what it does, and I wouldn't dream of writing anything even close to it on my best days, let alone trying to do it as an open source project. I know it doesn't sound like it because I'm a prick by nature, but I'm actually a huge fan of bokeh and I'm not being sarcastic.

Alas not that it matters but I don't believe there is any way for scatter to have the on hover effect of scaling size. I spent a couple of hours trying to do this to no avail with the provided API.

Thanks for the link. It makes a lot of sense that radius is measured in milliseconds and I should have made the connection. Usage error, my mistake.

[bryevdv]

Alas not that it matters but I don't believe there is any way for scatter to have the on hover effect of scaling size.

I literally posted a screenshot above demonstrates that it is possible (with 3.4+). Here's another, explicitly with scatter and size:

But if you would like to discuss usage questions, the community discourse is the correct place, not the tracker, as I already stated.

[u3Izx9ql7vW4]

That is a relevant clarification. I'm not in your shoes and frankly I don't do what you do not only because I can't, but I don't want to for the same reason you're thinking of leaving. But it's safe to say a lot of people use Bokeh for things which are a net positive in the world, and I'd be doing them a disservice if I didn't ask you to stay your course.

Not sure you care, but I thought you should know I wasn't just willy nilly trying to piss maintainers off, and I didn't know you wrote Bokeh. My frustration came from spending Friday + the entire weekend adding a border around a plot & trying to get a data point to pop on hover, and I was able to only do one of those things. And I did go through the documentation -- in fact it was the only thing I could do because I couldn't make sense of the code base, and a lot of the CSS related docs did not work for me.

I apologize.