Here is an update,
Kostja gave me access to his 2.95.3 binaries and that took care of the
template problem (so, I'll have to upgrade our minimum requirements
for gcc).... Unfortunately, now something else is going wrong, but it
is not clear whether my changes are causing it or merely revealing it.
I compiled the emulator with DEBUG_CHECK switched on, and this reveals
a curious problem in dictionary.{cc,hh}: Assert(oz_isFeature(a))
fails. Using gdb, I noticed that the DictNode being processed
contained 0x4f4f4f4f as the key, and coincidentally, this is the value
that we write into ozMalloc'ed blocks when DEBUG_CHECK is on.
Right now, I am looking at DictHashTable::resize() and "p /x *old"
shows that both the key and the value in the _old_ table are
0x4f4f4f4f. This makes no sense.
I could use some help.
(1) could someone else try it out and see if they have these problems
(2) could someone grant me access to a _fast_ solaris machine.
I have been using scheutz at sics, but the long recompilation
times make the debugging process really painful.
Cheers,
-- Denys Duchier - Équipe Calligramme - LORIA, Nancy, France - Please send submissions to hackers@mozart-oz.org and administriva mail to hackers-request@mozart-oz.org. The Mozart Oz web site is at http://www.mozart-oz.org/.