[tex-k] about typedef boolean

[Karl Berry]

    For example, Fabrice had problems compiling luatex because his boolean
    typedef definition did not match the one in web2c's kpathsea.

If you do #define HAVE_BOOLEAN before #include-ing kpathsea/whatever.h,
kpathsea/types.h will refrain from typedef'ing boolean.


Anyway, I looked at cpascal.h.  The stock web2c/cpascal.h already
#includes several kpathsea headers, so it's certainly going to get
kpathsea/types.h, so as things stand, it seems it would be useless to
define boolean there.  Can you explain how it will help?  (I'm not
trying to be difficult, just don't understand.)

Did you remove those #includes in your version, or something?  Maybe we
should split web2c/cpascal.h into the non-kpathsea-related stuff and the
other stuff?  What does your web2c look like?  Or you just living with
it as-is?  If you've made other changes, let's try to merge them.


Anyway, I see other dependencies of web2c on kpathsea (related to
infrastructure, not file searching) -- in fixwrites and splitup.  Plus,
I see Olaf copied a few functions into web2c/kps.c.  Did you already
untangle all that?  Seems like a lot of work for relatively little gain
to make a 100% independent web2c (not that I reject the idea in
principle).  Duplicating code seems like a bad idea, too -- Web2c was
never really standalone.  Hmm.

karl