transformations

[Jason Stover]

I have another question about transformations. Here is
the syntax:

data list list / v0 to v2.
begin data
 0.65377128  7.735648 -23.97588
-0.13087553  6.142625 -19.63854
 0.34880368  7.651430 -25.26557
 0.69249021  6.125125 -16.57090
-0.07368178  8.245789 -25.80001
-0.34404919  6.031540 -17.56743
 0.75981559  9.832291 -28.35977
-0.46958313  5.343832 -16.79548
-0.06108490  8.838262 -29.25689
 0.56154863  6.200189 -18.58219
end data
regression /variables=v0 v1 v2 /statistics defaults /dependent=v2 /method=enter 
/save=residuals.
descript all/stat=all/format=serial /save.
list.

The function subcommand_save() in regression.q adds a new variable to
the dictionary, but the transformation (regression_trns_proc()) it
adds is never executed.

On the other hand, the DESCRIPTIVES procedure in the syntax above also adds a
transformation, and that transformation is executed later. I read
definition of descriptives_trns_proc() and setup_z_trns(), and I don't
understand why the DESCRIPTIVES transformation is executed but the
REGRESSION transformation is not.

So am I missing something in the subcommand_save() or in
regression_trns_proc()? Or something else entirely?

-Jason