Petri Koistinen writes:

[...]

  Many of the transformations you made are, at best, negligible,
and, debatably, make the code less readable (like replacing tabulation
for aligning identifiers in blocks of definitions or declarations
by a single space etc.) But most important, have you run extensive
regression tests on each of those programs you have changed, making
sure that each one exhibits exactly the same behaviour before and
after your style conversion? There are subtle differences with type
promotion and call semantics in the presence or absence of prototypes,
with regard to old style and new style function definitions, and
the stuff you're converting is often from pre-ANSI days when certain
assumptions were made that perhaps do not hold anymore.  I would
like it if you also published the logs of the regression tests you
made (and the scripting of these tests aswell) to make sure the
utilities' behaviour did not change.  I'd hate to have a basic
utility fail just because of "style improvements".

--mkb