RE: [HACKERS] pg_dump problem

This looks like a place where the sprintf is being caught with too long a
string.  Although 487 bytes shouldn't be too long; perhaps the rest of
what's in there is causing sprintf to hooch.  The limit is supposed to be
around 1kB.  I'll try changing the fmt style string to a series of appends.
It's less efficient, but may work.

Thanks...


MikeA




-----Original Message-----
From: Patrick Welche,SCC,ext.35710,
To: pgsql-hackers@postgreSQL.org
Sent: 00/01/07 12:56
Subject: [HACKERS] pg_dump problem

% pg_dump -sv regression > foo
...
--  dumping out user-defined procedural languages 
--  dumping out user-defined functions 
Segmentation fault (core dumped)

#0  0x4810dc02 in vfprintf ()
#1  0x480c4963 in vsprintf ()
#2  0x48078e8a in appendPQExpBuffer (str=0x616e746f,    fmt=0x3d20656d <Address 0x3d20656d out of bounds>) at
pqexpbuffer.c:197
#3  0x6c732065 in ?? ()

% tail foo   end if;   return new;
end;
' LANGUAGE 'plpgsql';
CREATE FUNCTION "tg_iface_biu" ( ) RETURNS opaque AS '
declare   sname       text;   sysrec      record;
begin   select into sysrec * from

ie., it stops in mid flight, the line is
   select into sysrec * from system where name = new.sysname;

so it seems that the PQExpBuffer may well be to full ?

-s means dumpSchema(), getFuncs(), dumpFuncs(), dumpOneFunc()

pg_dump.c:dumpOneFunc():2346:
       appendPQExpBuffer(q, " ) RETURNS %s%s AS '%s' LANGUAGE '%s';\n",                       (finfo[i].retset) ? "
SETOF" : "",                       fmtId(findTypeByOid(tinfo, numTypes,
 
finfo[i].prorettype), false),                       func_def, func_lang);


so it cored while printing func_def ?! which is only 487 bytes long..
Rather confused.. Are any of you seeing this?

Cheers,

Patrick

************