\centerline{\bf Too many braces take a belting}
\smallskip
\noindent
For some good reason, which now escapes, I had cause recently to look 
something up in number 7 of this series; thus proving that it is fulfilling its 
destiny as a standard work of reference.  As I was browsing through my eye 
lighted on the review of a software package called Formulator.  The exact 
nature of this package is of no consequence to my remarks but its pertinent 
property is that it produces `software-written' \TeX\ code for, as the name 
suggests, mathematical formulas.  The reviewer points out that it has a 
number of eccentricities in its use of braces: for example, a tendency to
surround  many subformulas with unnecessary braces but to omit them in some
places  where they are essential.  The problems caused by the latter are
explained  in some detail, as is one change in the expected typography caused
by the  insertion of an empty pair of braces.  The rest are categorized as `No
bad  thing'.

I would argue that all such extra braces are not simply superfluous but are 
possible causes of incorrect typesetting of a nature which is particularly 
disturbing in a system which is presumably aimed at a class of user who is 
less likely than some to notice these errors since they largely affect those 
details of spacing in formulas of which most readers and producers of 
mathematical texts are blissfully unaware.

The cause of the errors occasioned by these braces is the fact that any 
subformula enclosed in braces (even an empty one) is an atom of type 
(flavour) Ord (see p.\ 291 of {\sl The \TeX book}) whatever its contents may
be.   Thus, choosing examples at random from the review: |{\left( ... \right)}| 
appears to the neighbouring atoms to be an Ord instead of its correct 
identity which is Inner; |{\int ...}|, |{\sum ...}| and |{\sin ...}| all change
from Op to  Ord.  A glance at the table on p.\ 270 of {\sl The \TeX book\/} will
reveal the havoc  which such changes can wreak to the subtleties of spacing
revealed therein.

Thus I am led to the conclusion that either the code-writing part of this 
software contains a fundamental design flaw or this software is not intended 
to produce high-quality \TeX-code and is therefore not intended for use in 
high-quality typesetting software.

Of course, it is just possible that it is merely another piece of evidence in 
favour of the thesis that the \wysiwyg\slash Graphical Interface\slash {\sc
wimp} paradigm  is fundamentally at odds with high-quality anything.

\rightline{\sl Chris Rowley}