memory leak in regression

[Ben Pfaff]

valgrind reports a small memory leak in the regression routines:

==29398== 
==29398== 16 bytes in 2 blocks are definitely lost in loss record 2 of 7
==29398==    at 0x401C422: malloc (vg_replace_malloc.c:149)
==29398==    by 0x80E62B6: xnmalloc (xmalloc.c:49)
==29398==    by 0x80C1FF1: pspp_linreg_coeff_init (coefficient.c:73)
==29398==    by 0x8091F90: run_regression (regression.q:1206)
==29398==    by 0x80D2F05: multipass_split_output (procedure.c:640)
==29398==    by 0x80D2F64: multipass_split_end_func (procedure.c:631)
==29398==    by 0x80D3A3D: internal_procedure (procedure.c:229)
==29398==    by 0x80D3D82: multipass_procedure_with_splits (procedure.c:594)
==29398==    by 0x80915ED: cmd_regression (regression.q:934)
==29398==    by 0x804FA32: cmd_parse (command.c:186)
==29398==    by 0x804D697: main (main.c:137)
==29398== 
==29398== 
==29398== 132 (60 direct, 72 indirect) bytes in 3 blocks are definitely lost in 
loss record 5 of 7
==29398==    at 0x401C422: malloc (vg_replace_malloc.c:149)
==29398==    by 0x41E1C60: gsl_vector_alloc (in /usr/lib/libgsl.so.0.7.0)
==29398==    by 0x80C25B7: pspp_linreg_cache_alloc (linreg.c:150)
==29398==    by 0x8091E0E: run_regression (regression.q:1147)
==29398==    by 0x80D2F05: multipass_split_output (procedure.c:640)
==29398==    by 0x80D2F64: multipass_split_end_func (procedure.c:631)
==29398==    by 0x80D3A3D: internal_procedure (procedure.c:229)
==29398==    by 0x80D3D82: multipass_procedure_with_splits (procedure.c:594)
==29398==    by 0x80915ED: cmd_regression (regression.q:934)
==29398==    by 0x804FA32: cmd_parse (command.c:186)
==29398==    by 0x804D697: main (main.c:137)
PASS: tests/command/regression.sh

-- 
Ben Pfaff 
email: address@hidden
web: http://benpfaff.org