Re: "type 3 sums of squares to GLM" commit test failures

[John Darrington]

Yes.   This is because of the changes in categoricals.c

For covariance matrices, the number of columns associated with a categorical 
variable needs to be one less than the actual number of distinct values in that 
variable.  Whereas other uses require the full number.

I'm mulling over the various ways in which we can fix it to work for both use
cases.  It shouldn't be hard.  But I'm worried that the interface to 
categoricals
and/or covariance will become unwieldy if we don't be a little careful.

J'

On Sat, Jun 25, 2011 at 07:27:08PM -0700, Ben Pfaff wrote:
     The "type 3 sums of squares to GLM" commit led to a number of
     test failures:
     
     ## ------------------------ ##
     ## Summary of the failures. ##
     ## ------------------------ ##
     Failed tests:
     GNU PSPP 0.7.8-g840f7b test suite test groups:
     
      NUM: FILE-NAME:LINE     TEST-GROUP-NAME
           KEYWORDS
     
      424: oneway.at:3        ONEWAY basic operation
      425: oneway.at:72       ONEWAY with splits
      426: oneway.at:179      ONEWAY with missing values
      427: oneway.at:284      ONEWAY descriptives subcommand
      429: oneway.at:378      ONEWAY multiple variables
      430: oneway.at:484      ONEWAY vs. weights
     
     Details in the log here:
     http://pspp.benpfaff.org/~blp/pspp-master/20110625030504/testsuite.log
     
     Thanks,
     
     Ben.
     -- 
     Ben Pfaff 
     http://benpfaff.org
     
     _______________________________________________
     pspp-dev mailing list
     address@hidden
     https://lists.gnu.org/mailman/listinfo/pspp-dev

-- 
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.

signature.asc
Description: Digital signature