Cannot use non-html5 equations

[ianhbell]

First off, excellent work with this package. I am able to convert a rather complex document to html5 with lots of equations and images. Ultimately my target is to get back to a Word document from a latex document. I think I am not far off.

That said, latexml always borks when I try to tell it to use svg or png equations (with --mathsvg or --mathimages respectively). The HTML5 equations will obviously be no good in Word.

I always get this error:

Fatal:perl:die Perl died
	Postprocessing LaTeXML::Post::MathML::Presentation Paper.html
	Wide character in subroutine entry at /usr/share/perl5/LaTeXML/Post.pm line 1188.

It would be nicer if the error would not kill LateXML, rather it would just skip that equation.

[ianhbell]

It turns out the offending character was a en-dash from the cp1252 character set (as in not the ASCII dash). This bad guy: http://unicode-table.com/en/2013/

[dginev]

Was that character in the original input document, or it was created internally during math parsing?

Very valid bug report here, will be fixed.

[ianhbell]

It was in the original LaTeX document. Actually, I had converted a Word document to LaTeX by copy-paste (not the most elegant solution, but the fine-tuning was simple enough). I think that's how the character ended up there. The en-dash are a real pain from the copy-paste standpoint.

[dginev]

That said, it shouldn't really break latexml, which is unicode-aware. Could you share the formula that had the unicode dash with us? Maybe we can use it to reproduce and find an elegant fix.

[ianhbell]

Perhaps this would do it (it was actually an inline equation, but I think the problem is the same):

$s_{\rm res}(T,\rho) = s(T,\rho) – s^0(T,\rho)$

[brucemiller]

hmm, it seems that tied hashes don't like to get utf passed to them? I've patched that up and it runs without error (or even warning). However, I wonder that it gives what you want: LaTeX (even outside of LaTeXML) essentially ignores the en-dash; perhaps it was intended to be a minus?