On 2020/05/02 07:37:31, hanwenn wrote: > On 2020/05/02 07:34:29, barrykp wrote: > > On Sat, May 02, 2020 at 12:25:15AM -0700, mailto:hanw...@gmail.com wrote: > > > please don't submit; I'm rearranging this file completely. > > > > > > Can you add your regression test + instructions on how to reproduce the > > > problem? > > > > I did add a regression test in the first patch set, but it breaks > > testing. It seems we have no way to include the SVG backend in > > regression tests (I can see one other example that tries to, using > > inkscape, but it was disabled in 2008). I had to remove the regression > > test in the second patchset in order to not break make test/check/etc. > > Do you have any suggestion about how to deal with it? > > > > The issue in the tracker shows how to reproduce the problem - do you > > want that added to the code somewhere? > > > > > > > > https://codereview.appspot.com/582010043/ > > the .ly in the tracker is sufficient. Thanks.
I don't completely understand, though: if we put the "utf-8-string" directly into the SVG output, the SVG browser might make other font choices, making the outline potentially incorrect. Is this a problem in practice? https://codereview.appspot.com/582010043/