TeX and CSS treat line heights in fundamentally different ways. In
TeX, every character is treated as a box of its precise height and
depth; the line height (\baselineskip) applies after characters have
been assembled into lines. In CSS, in contrast, every character
creates a "line box" corresponding to the accompanying font. When
characters of different fonts and sizes are placed into the same
span, the resulting line box contains the line boxes of all children.
This is unfortunate because, for example, we want `\frac{1}{2}` to
behave in vertical spacing contexts like it is exactly as tall and
deep as the visible fraction (which is the TeX behavior). Given CSS
constraints, though, in most contexts the fraction has extra vertical
space: the line boxes for the numerator and denominator create
padding. For small boxes, this isn't so bad. To really see the
problem put a tall rule in the denominator of a fraction, or check
out the VerticalSpacing screenshotter test, which has way more space
than it should.
Solving this problem in CSS is difficult. There is no easy way to get
rid of the extra line boxes.
But there is *a* way, namely tables. A table-cell with vertical-align
top, bottom, or middle is ignored for the purposes of line height
calculation.
So in this commit, makeVList puts its contents into a
vertical-align:bottom table-cell (to clear unwanted line boxes), with
an extra row used to represent depth.
Many Chrome screenshotter tests change. This is because Chrome rounds
table dimensions to integral numbers of pixels, while it uses
sub-pixel positioning for non-table displayed tabs. That makes many
vlists a fraction of a pixel wider than they used to be.
* Fix interaction between styles and sizes by implementing styles as sizes.
Rather than having both `textstyle` CSS classes and `size5` CSS classes
affect the font size (and step on each other), implement sizes more the
way TeX does: a command like `\displaystyle` changes the current size.
This is actually a simplification, since now only `size` affects the size.
Simplifies CSS and computation. Many screenshotter tests change; they
change to be more like TeX. For instance, `\sqrt` fixes some
discrepancies in size treatment.
Also:
Remove the `Options.withX()` methods in favor of `.havingX()`, which
might return the same `options`.
Remove `Style.cls()` and `Style.reset()`.
Remove `Options.reset()`. You should never modify an `Options`; they
should change only by the `havingX()` methods.
* Implement TeX sizing for scriptsize/scriptscriptsize.
At every size level. Also make the sizes match TeX to the last decimal.
* Review comments.
The combination of jspngopt and pako should eliminate possible causes for
different PNG encodings, although the core reason for #325 remains unknown.
Pako has poorer compression rates than native libz, but optimization can
counter that effect, and actually reduce the size of the screenshots.
The screenshots for LimitControls and UnsupportedCmds on Firefox used to
exhibit subpixel rendering before, for reasons unknown. The regenerated
versions don't exhibit this. See #324 for a discussion.
Summary:
Create our own screenshotting script which takes screenshots. This
improves over huxley for a couple reasons:
- It makes the screenshots the correct size (for some reason, huxley struggles
with this).
- Its configuration matches more with what we want (we don't need multiple
screenshots or interaction, we just want a single static shot)
- It runs faster
I also changed the docs to reflect this change.
Test Plan:
- Make sure all of the tests that were in the Huxleyfile are now in ss_data.json
- Run the screenshotter docker
- Make sure all of the images look reasonable and don't change (except
sometimes the Lap test, which has some strange pixel-positioning
differences...)
Reviewers: kevinb
Reviewed By: kevinb
Differential Revision: https://phabricator.khanacademy.org/D16731
