Hi Everyone,

I’ve recently been wondering why it is that for sensibly sized images in a
LaTeX-export-oriented document I need to do both:
┌────
│ #+attr_org: :width 400
│ #+attr_latex: :width 0.4\linewidth
└────

When in HTML, just
┌────
│ #+attr_html: :width 400px
└────
is fine.

This has lead me to have a look at `org-display-inline-images', and I’ve realised
that the `#+attr_latex' width of `0.4\linewidth' is actually picked up, and
interpreted as a width of `0.4' pixels. I think it would make much more sense for
fractional values between zero and one to be interpreted as that portion of the
text width in the buffer. On second thoughts, given that the document width can
be slightly larger than the text width, perhaps an upper bound just a bit higher
— say 2, could be better.

I’ve prepared a patch which implements this logic, by converting extracted
widths that are:
⁃ floats, and
⁃ within the range [0,2]
and sizes them as that proportion of the text width in the buffer, which is
determined by checking
1. `visual-fill-column-width', when that package is installed and the mode active
2. `fill-column', when auto fill is active
3. `(window-text-width)', if neither of the above two cases hold

Please let me know what you think 🙂.

All the best,
Timothy