[with patch, positive review] latex issues

[jhpalmieri]

Here are several LaTeX issues:

• because of how Python works (I think), if you set T = type(identity_matrix(3)), then T has all of the methods of an identity matrix. In particular, if you run latex(T), the code calls hasattr(T, '_latex_'), and this returns True because hasattr(identity_matrix(3), '_latex_') is True. But then T._latex_() produces a TypeError. This is the cause of the error reported  here.

Solution: catch TypeErrors when calling T._latex_() in this sort of situation.

• In the notebook, try

  %latex 
  $\sage{type(35)}$ 
  

In this case, Sage typesets the string <type 'sage.rings.integer.Integer'>, but the < and > signs get converted into an upside-down exclamation point and question mark.

Solution: typeset strings differently, using \texttt instead of \text.

• Click the "Typeset" button and try

  type(35) 
  

In this case, jsMath kicks in and tries to typeset \text{<type 'sage.rings.integer.Integer'>}, but the symbols < and > confuse jsMath -- it thinks they're part of an html command. As a result, there is *no* output at all.

Solution: for typesetting strings in jsMath, replace \texttt with \hbox.

• This comes from a Sage doctest:

  sage: R.<x,y>=QQbar[] 
  sage: latex(-x^2-y+1) 
  -x^{2} - y + \text{1} 
  

The \text{1} should not be there.

Solution: The \text{1} appears because the element 1 in R has no _latex_ method, so it gets converted to a string, when then gets typeset by enclosing it in \text. So test strings: if they consist only of digits, just return the string. If they contain anything else, enclose in \texttt, as mentioned above.

A patch will follow soon.

[jhpalmieri]

I'm attaching a patch which implements all of the solutions discussed above. When applied to version 4.1.2.alpha4, this passes all doctests on sage.math.

If you have questions about the design decision, say the appearance of strings typeset in \texttt vs. \text, well, for what it's worth, I tried to discuss some of these issues on  sage-devel a few months ago, but there was essentially no response. With this version, typesetting Python strings via LaTeX will produce different looking output, but (a) I like the new look, and (b) this seemed like the best way to deal with symbols like <, >, and _.

If necessary, we can split this into several tickets, because the first issue (_latex_ method for types) is separate from the others.

[timdumol]

Applies well. Things look good here. Positive review.

[mhansen]

I had to rebase the patch to get it to apply to my current tree.

[mpatel]

If I evaluate

html('$\CDF$')

in the notebook, jsMath complains

Unknown control sequence '\texttt'

Is this because sage.misc.latex_macros.sage_jsmath_macros contains

jsMath.Macro('CDF','\\\\texttt{Complex Double Field}');

?

[jhpalmieri]

In Sage 4.2.alpha0, if I evaluate

html('$\CDF$')

in the notebook, I get the message

unknown control sequence '\CDF'

I can't reproduce the error message about \texttt, though.

[mpatel]

The sagenb spkg in 4.2.alpha0 and all more recent versions (to date) do not load the jsMath macros. I'm working to fix this. To see the \texttt message, I did

sage: from sage.misc.latex_macros import sage_jsmath_macros
sage: for m in sage_jsmath_macros: print m

and inserted the definitions

jsMath.Macro('ZZ','\\Bold{Z}')
jsMath.Macro('RR','\\Bold{R}')
jsMath.Macro('CC','\\Bold{C}')
jsMath.Macro('QQ','\\Bold{Q}')
jsMath.Macro('QQbar','\\overline{\\QQ}')
jsMath.Macro('GF','\\Bold{F}_{#1}',1)
jsMath.Macro('Zp','\\ZZ_{#1}',1)
jsMath.Macro('Qp','\\QQ_{#1}',1)
jsMath.Macro('Zmod','\\ZZ/#1\\ZZ',1)
jsMath.Macro('CDF','\\texttt{Complex Double Field}')
jsMath.Macro('CIF','\\Bold{C}')
jsMath.Macro('CLF','\\Bold{C}')
jsMath.Macro('RDF','\\Bold{R}')
jsMath.Macro('RIF','\\Bold{I} \\Bold{R}')
jsMath.Macro('RLF','\\Bold{R}')
jsMath.Macro('CFF','\\Bold{CFF}')
jsMath.Macro('Bold','\\mathbf{#1}',1)

into

SAGE_ROOT/local/lib/python/site-packages/sagenb/data/templates/notebook/head.tmpl

by hand.

[jhpalmieri]

You know, the right choice might be to remove 'CDF' from the list of macros. After all, the whole point of the list is to be able to use \RR in a docstring and have it look nice. Using \CDF in a docstring won't look very good with '
texttt{Complex Double Field}' or any variant on it. The only place I see '\CDF' in the Sage library is in sage.misc.latex, where it comes from sage_jsmath_macros. (I did search_src('\\\\CDF').) So I think we should delete it: it was a mistake of mine to include it in the first place.

[mpatel]

Removal sounds good, but in case people use it outside the Sage library, should we ask on sage-*?