[Draft] Adding support for smooth paths

[francoisthire]

Solve #33

Context

This PR adds two functions smooth_ccurve and smooth_qcurve which allow to define smooth curves. One of the control points is defined depending on the previous curve.

The function smooth_ccurve corresponds to s and S SVG path commands, while smooth_qcurve corresponds to t and T SVG path commands. More details here

TODO:

• [x] Check the PR compiles correctly (I did not find brzo on opam. I will pin it later)
• [x] Add some tests
• [x] Update CHANGES.md

Questions to the reviewers

• Which kind of test should be added? Some few unit tests?

[dbuenzli]

[ ] Check the PR compiles correctly (I did not find brzo on opam. I will pin it later)

The official build system is still topkg so simply opam install topkg-care and do topkg build.

[dbuenzli]

• Which kind of test should be added?

You can add a sample image in the database, named path-smooth, that shows use of the combinators see for example path-cubics (src).

You can then have a look at the results by running for example rsvg.native. See here on how to build the tests. One day I hope to have a raster renderer and expect test the renders with pixeldrift. But meanwhile that should do.

[francoisthire]

I followed your instructions:

• The project compiles correctly with topkg
• I have added a test ensuring smooth paths work correctly (I was not very inspired and used the classical quarter circle approximation). The code is a bit ugly but should be easy to understand.

Here is the result:

The implicit control point is colored in red.

[francoisthire]

I have updated the CHANGES file. Let me know if there is anything more that I can do.

[dbuenzli]

Thanks!

I squashed your patches into 3b3cea5 and 2d77ab2 and cosmetically massaged them in 33e1952.

Using trigonometry to compute point reflections felt rather heavy weight (and more expensive) so in 1e690d3 I replaced computations by simple linear algrebra following this diagram:

[francoisthire]

Perfect, thanks!