
* change font configuration and implement new mathvariant
* match characters with variants

MathML Implementation

* leftoverlap, rightoverlap in side attribute of mtable
* what if minlabelspacing is a % ?
* axis in rowalign
* table layout is not precise sometimes (e.g. when there is a fit
  table inside a auto column)

Optimizations

* optimize scrolling, it seems to work fine now (see gtkmathview.cc) but
  it has to be tested more heavily
* FontifiedString is no more necessary (not true, until monospace in LaTeX
  is not properly configured, se a mIxEd case word)

Major improvements

* definitely use iostream
* remove String class and use a more efficient implementation, based on the
  consideration that the only multi-char strings are only made of
  plain characters (or better, alnum characters), and that all the other
  characters (and stretchable chars) can be store with no significant waste
  of space in a 32bit field
* kerning is NOT available when exporting PS, the whole issue of tokens
  must be reconsidered and reimplemented

Minor improvements

? making script elements breakable
* more comments in the source code
? SetLastBreakability more consistent, what happens if several breakability
  are specified? Take the weakest one?
* hash table for keywords
* memory leaks
* if there is some trailing garbage in an attribute value then
  a warning should be displayed
* using a glyph for the / in bevelled fractions

Configuration

* configurable parameters
  (spacings for roots, default font attributes)
* support for STIX fonts in font-configuration.xml
* configurable dash length for dashed lines in PostScript

Bugs

* Some chars (mustache) are compound only, they do not have a "simple" form
* problem with color and PostScript generated files
* in font-configuration.xml, the font map for monospace is not complete
* what if there is a horizontal stretchy operator with minsize specified in a mrow?
* findStretchyOperator does not always work properly, please check
* MediumSpace is now U+205f, but the specification is not consistent. The table
  reports U+205f, but the chapter reports U+2005
* operators strechying within tables buggy. Sometimes the table remains "larger"
  than needed just because there were a strecthy operator
* it seems that `gtk_math_view_set_font_size' is called twice when a new default
  font size is selected. Now I've patched this by making a check and preventing
  re-doing the layout if the font turns out to be equal
* what if one wants a large operator which is not in a stretchy position or
  it has strethcy="false"?
* cannot select phantom elements
* check examples in problems under solaris, there might be a weird bounding
  box
* check length of dashes when rendering to postscript

