ancient LaTeX approach of using "mathematical bold/italic" ranges

Actually, LaTeX predates Unicode so it looks up font files of different styles. WpfMath does this: they have different fonts for the same codepoint that contain different styles. However, iosMath uses Latin Modern Math by default which combines all these fonts together into the mathematical glyph ranges, so we can use Unicode from top to bottom.

Actually, LaTeX predates Unicode so it looks up font files of different styles. WpfMath does this: they have different fonts for the same codepoint that contain different styles. However, iosMath uses Latin Modern Math by default which combines all these fonts together into the mathematical glyph ranges, so we can use Unicode from top to bottom.

Interesting. WpfMath then ends up with a more convenient approach with fewer special cases but I think this PR can go ahead without changing that. We'll just need to preserve some degree of "font changing" for TextAtoms.

Can you understand the test failures, or do you remember why there was a DoNothingFontChanger in the tests (before the changes in the cleanup so far) @Happypig375 ?

It would be nice to remove test failures before pushing the real work. I can't see any stated meaning of the tests so cannot evaluate. Presumably the tests were not testing actual behavior (since they required DoNothingFontChanger to pass) and so removing DoNothingFontChanger is a matter of hygiene so that the tests test real behaviour.

Can you understand the test failures, or do you remember why there was a DoNothingFontChanger in the tests (before the changes in the cleanup so far) @Happypig375 ?

It's just a mock for UnicodeFontChanger in TypesettingContext. It's a leftover from @verybadcat.

I can't see any stated meaning of the tests so cannot evaluate.

Yes, there are some tests that test the mocks themselves.

The test failures are related to using UnicodeFontChanger instead of DoNothingFontChanger. Since the text now have characters outside of the BMP, the difference in UTF16 length leads to different IndexRanges. Might have to check iosMath on its behaviour.

This anomaly can be traced all the way back here:

Definitely not correct. CSharpMath.Editor uses Ranges as input to constructing MathListIndexes. Ranges must represent locations of atoms and using text length instead is a logic error.

You mean UnicodeFontChangerTests?

UnicodeFontChanger/Range tests.

Range tests are important since they are part of the atom output. Not part of the intermediate steps in creating atoms.

Alright, all tests now pass.

@charlesroddie I've changed UnicodeFontChangerTests so that they also test LaTeX->Display conversion instead of only testing UnicodeFontChanger.

@charlesroddie 2 outstanding reviews should be resolved for this to be mergable

@Happypig375 Are the two things to be resolved the questions I asked about the tests?

Can this be merged?

There are 2 unresolved conversations

You understand the remaining conversation so can you mark if you consider it resolved?

Codecov Report

Merging #156 into master will decrease coverage by 24.98%.
The diff coverage is 98.34%.

@@             Coverage Diff             @@
##           master     #156       +/-   ##
===========================================
- Coverage   86.81%   61.82%   -24.99%     
===========================================
  Files         157      254       +97     
  Lines       11275    20793     +9518     
===========================================
+ Hits         9788    12856     +3068     
- Misses       1487     7937     +6450     

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 789b861...20f03b9. Read the comment docs.

Charlie and whomever pay attention to the master here and I am here to tell you both that the house is due to be cleaned of the filth and chaos that you have done. You have many incorrect coding and you're not even close to be verified. The mit license doesn't belong to you at all it's mine. Your hatred is a terrible thing to have and I didn't do anything wrong for you to do this.

@Atlantis433 Do you have some problems that are unresolved? Especially the "The mit license doesn't belong to you at all it's mine.". Please write in a calm and understandable way so that we can resolve these issues separately.