🚀 go-pugleaf

etc/MACHINES mentions that it is unknown whether emacs will work
on the MIPS based Pyramid systems.  It _almost_ does...

Sorry for the lack of detail, this system is at a site that has
no outside links.  Also I have not yet seen any documentation
for this system so can't be more specific yet.


Anyway, for the interested, I used

s-usg5-4.h, m-pyramid.h and added these to config.h

#define SYSTEM_MALLOC
#define HAVE_ALLOCA
#define CANNOT_DUMP

The first two reduced the frequencey of memory faults and bus
errors. 

Emacs compiled with only ONE warniing about garbage after a
pre-processor directive.

Use make CPP=/usr/ccs/lib/cpp as cc -E gets it all wrong.

(producess muck like $ { XXXX } vs. ${XXX})

At this point, xemacs works pretty well, but re_search etc does
not work.

I noticed this when dired mode claimed there were no files to be
found.  Debugging dired.el showed that it was a RE search
failing.

Doing the same search manually showed (in a dired buffer).

"\\(Oct\\|Mar\\)[ ]"

matches ok but

"\\(Oct\\|Mar\\)[ ]+"

or

"\\(Oct\\|Mar\\)[ ]*"

failed.

In fact I still get core dumps from re_seach() causing recursive
calls to re_match_2().

I'm sure there will be something simple.

I've looked at the regex.c code briefly and can't see any reason
for it not to work.   I'm guessing that the problem is caused by
recursion (all of the core files I've looked at showed it).

Please e-mail any suggestions, as I have little time just now
for scanning news.

Simon J. Gerraty        <sjg@zen.void.oz.au>

#include <disclaimer>   /* imagine something _very_ witty here */