I now have a few cycles to look at the portable dumper bug a little more,
and
figured out that (setq-default ) on an opaque type doesn't work properly
after
the dumper reloads it into memory. It appears to apply to
SYMVAL_BOOLEAN_FORWARD variables (defined as DEFVAR_BOOLEAN)
that are buffer local. Don't know yet if it's more pervasive than that.
Who's the portable dump expert, and are there plans to fix this anytime
soon?
If not, I'd like to ask a few questions and see if I can fix it myself...
but don't
know if the list wants to see all that traffic, so could correspond with
them
offline if need be.
It may be as simple as making sure the opaque variables get into the
reloc_table so they get handled by the call to pdump_reorganize_hash_table.
I'm not sure yet though.
dumper.c, Line 1040 is the last time the variable gets "touched" if
my debugger watchpoint doesn't lie.
/* Put back the opaques */
for (i=0; i<((dump_header *)pdump_start)->nb_opaquedmp; i++)
{
struct pdump_dumpopaqueinfo di = PDUMP_READ (p, struct
pdump_dumpopaqueinfo);
memcpy (di.data, p, di.size);
p += di.size;
}
Thanks, Troy