Re: regression procedure

[John Darrington]

On Sun, May 15, 2005 at 07:28:32PM +0000, Jason Stover wrote:

     It ran with no trouble, so I added more code.
     www.sakla.net/regression.q shows the latest version. With this syntax
     file:
     
        data list / v0 to v2 1-9.
        begin data.
         12 12 89
         34 12 80
         56 12 77
        end data.
        regression /variables v0 v1 v2 /dependent v0 /method=enter.
     
     pspp calls run_regression(...) with no trouble. This function exits,
     then cmd_regression() exits normally. Then pspp is in parse_script ()
     at main.c:101, and handle_error(execute_command()) never returns. This
     problem never occurs with old_regression.q. Does anyone know what's
     happening here?  

Our trusty friend valgrind shows:

==19841== Invalid free() / delete / delete[]
==19841==    at 0x1B904B04: free (vg_replace_malloc.c:152)
==19841==    by 0x8071181: cmd_regression (regression.q:56)
==19841==    by 0x807D306: cmd_parse (command.c:255)
==19841==    by 0x809EA6B: execute_command (main.c:134)
==19841==    by 0x809EA1C: parse_script (main.c:104)
==19841==    by 0x809E9B8: main (main.c:91)

So I deleted line regression.q and everything ran fine.  I didn't see
any output, but I guess you haven't implemented that bit yet.
     
     
     And does regress.q have the right look for a new
     procedure?  (Aside from the fact that it doesn't work yet.)
     
It looks good to me.


J'

-- 
PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See http://pgp.mit.edu or any PGP keyserver for public key.

pgpfd6BPjIi6O.pgp
Description: PGP signature