reduce-algebra/generic/casefold

casefold.c


This is a perhaps silly utility that can be used to convert all parts of a
file that are not in strings and not parts of comments to lower case. It has
a list of exceptions for words that should NOT be folded to lower case.

Its purpose is to make it possible to finalize the adaptation of Reduce
from its original upper-case-only world to one where eventually we can
make the system case sensitive.

Anybody offended by the changes it has made to their code may either edit
their files back to the state they want manually or they may add the mixed
case words that matter to them to the list here and use this to apply
changes systematically. In most cases where package maintainers had used
mixed case at all I found multiple instances where they had not been
consistent in its use.

A harder issue is competing modules that use the same name but where there
is disagreement as to the best case to use. Especially for local variables
people ought really to be able to use whatever capitalization they like
within their own package - but for simplicity I will apply global adjustments.

The change I made that perhaps led me to need to alter most odd files was
making the fully spelt out word COMMENT always appear in upper case in
source files. This led to the need to adjust many log files. Note that
"comment" and "Comment" will still be treated as introducing a comment
if Reduce input has case-folding enabled. I believed that the bold word
COMMENT was a better thing to standardise on than its lower case varient.

When initially building this I will avoid processing anything in the redlog
directory even though there are inconsistencies there so that the relevant
authors can fix things up be hand! But otherwise I go something like

for x in packages/*/*.red packages/*/*.tst
do
./casefold $x
done

to standardise the bulk of Reduce. I then need to check and review all
the .rlg files where they echo input that has been subjec to case change.
I prefer to edit those individually as that helps ensure I have really
considred log-file changes...



Arthur Norman. March 2015