The files I have changed can be found by looking for my initials:
fgrep JNP *on Unix. All changes are commented. (I have not, at this stage, put any files into the project library, but have left them for Steve.)
Briefly, the files I have changed are:
AUTORUN.LSP. I have generalised the index-lookup routines
so they can be used on more than one index. There is now an
access function
D-LookUpInIndex (word fn index)This looks for
word in verb/index/, and applies fn to the
result, passing its result back.
ENTRCONV. This contains some routines called when compiling
lexical entries which we have modified. The lexicon-building routine now
constructs both the normal index, and the IFI. I have changed it to
convert the surface forms and roots to strings so that when read back,
they won't occupy symbol-table space. It also downcases the roots.
Modify this file to change the format of compiled entries, or the way
the index is linked to the entry file via file-positions.
KEYWORDS. This contains the file extension for the IFI
file.
MAKELEX.LSP. This reads and parses the lexicon, and writes
it out at the end. I made a minor change to fix a bug in the output
routine. I also changed the way the incore-flag was handled: as
previously implemented, it always got reinitialised before every call to
D-MakeLexicon, and hence had no effect.
MALOAD.LSP. This is the top-level MA file which loads all
the others. I have made it loads MORPHGEN.LSP too. Also, I have
fixed the implode
macro, at the suggestion of Alan Black. There was a
bug which stopped D-MorphemeConcat from supressing zeros.
MORPHGEN.LSP. This is the morphological generator.
PAROUTS. Changed how the incore flag was handled.
SMAFUNCS.LSP. This loads the word-grammar rules. I have
modified it to load the generational ones too. I have also changed the
incore flag again.
SMKWGRAM.LSP. I have modified this to compile the
generational word-grammar rules.