do we still need cairo and svg for some (but not all) images in the docs?

[newville]

Can we drop the use of cairo and svg images in the docs? Or, should these be used more widely?

[reneeotten]

@newville I would certainly prefer to keep those as that will result in high-quality, scalable images in the documentation. As far as I know the only place where SVG images are not used is in the Examples Gallery. I would like to use it there as well but back in the days when I spent quite some time on this, there was no (easy) way to accomplish that. I can take another look and see if sphinx-gallery can be convinced to generate/use SVG files as well.

[newville]

@reneeotten OK. I guess I don't see a huge difference in image quality. And it seems like it is used in only a few cases. Should it be used more consistently?

[reneeotten]

as I said above, I think the only place where we don't use SVG images is in the Example Gallery. Sure, we could use it there as well if one can figure out how to do so.

[newville]

@reneeotten is it guaranteed that jupyter-execute sections of a document share a configuration? I'll admit that I find myself continually baffled by Jupyter vs Python.

I think the docs could use some work, but I'm afraid that I don't really understand them.

[reneeotten]

I suppose these statements share a configuration, but I don't know in what context you ask this. Are we still talking just about png vs. svg images? That should controlled by the "sphinx-gallery" package.

Also, can you elaborate on what you mean by "the docs could use some work"?

[newville]

@reneeotten Thanks -- I meant in the overall docs. Like, if I want to add or change examples, it seems I should be using jupyter-execute instead of inserting images. But when doing that, the jupyter-specific-non-Python-configuration command to "use SVG" is sometimes required to be explicit and (apparently) sometimes implicit. Apparently Jupyter is caching state information (like, a full "session") between jupyter-execute directives/sections....

For "what to add to the docs", I think it might be helpful to start with a more Model-based introduction. The main docs are pretty formally based on Parameters and Minimize, and a "gentle introduction to curve-fitting with Model" might be overdue.