legend entries just in {} instead of $$?

[matlab2tikz]

legend entries just in {} instead of $$?

[CapsLocker]

changing the interpreter on the legend fixes this one I think:

lh = legend('show','Location','north west');
set(lh,'Interpreter','latex')

regards

[burkart]

I wouldn't recommend that workaround unless you know what you're doing.

My goal in rewriting the string handling was to get as close a result in the matlab2tikz output to what you see in MATLAB itself. This means that m2t is "bug-compatible" with MATLAB, for instance when it comes to MATLAB interpreting "\alpha \beta" as two greek symbols with a space in between even though a proper TeX interpreter would omit the space as it is the termination of the control sequence. (If you indeed wanted to have a space, proper TeX markup would require either 2 spaces between the greek letters or the sequence "\alpha{} \beta".)

Now, the "latex" interpreter as understood by MATLAB is more true to proper (La)TeX syntax as opposed to MATLAB's "tex" interpreter -- or at least that's what's in the documentation. Judging by the examples on the MathWorks website, text that is to be interpreted by the "latex" interpreter typically, if not always, is put inside an equation environment "$$...$$". What m2t does in the function prettyPrint() is to replace both the two dollar signs at the start and at the end of the string with single dollar signs to use TeX math mode. Since the replacement uses regexprep() there's no replacement whatsoever taking place if the matching criterion "$$...$$" isn't fulfilled.

That's why your workaround works, assuming the legend entries aren't encapsulated in "$$...$$". However, MATLAB might treat them very oddly which brings us back to the original point: Make your text elements like legend entries look the way you want them to in MATLAB first. If the output generated by m2t then deviates from that, please file a bug report. That way it shouldn't be necessary to do any formatting specifically for either MATLAB or matlab2tikz -- which was the goal in rewriting the text handling to begin with.

[burkart]

First of all, the solution for the original submitter might be more or less what CapsLocker suggested, except that the "interpreter" property shouldn't be set to "latex" but to "none".

Additionally, what was discussed in conjunction with github issue #64 might solve the problem, i.e. the new m2t parameter "parseStrings". Therefore I think this issue should be closed.