rfc - Perl_ck_op_sanity

folks,

It turns out that there is no Perl_ck_ routine that checks *all* 
constructed ops.

ISTM that we should have one.
If nothing else, it gives us a place to put invariants,
and start to understand the science below the magic.

These asserts work (theres some overlap, but minimalism seems 
counterproductive)

the last one feels clumsy - it special-cases mapstart vs grepstart;
is there something extra sneaky going on here (

What other invariants are possible/informative/worthwhile ?


diff -ruNp -X exclude-diffs ../bleadperl/op.c ckop/op.c
--- ../bleadperl/op.c   2008-03-10 15:43:52.000000000 -0600
+++ ckop/op.c   2008-03-13 14:52:07.000000000 -0600
@@ -6266,10 +6266,28 @@ Perl_newSVREF(pTHX_ OP *o)
 /* Check routines. See the comments at the top of this file for details
  * on when these are called */
 
+static
+void
+Perl_ck_opany(pTHX_ OP const * const o)
+{
+    int i = 0;
+
+    /* borrow one for now */
+    PERL_ARGS_ASSERT_CK_ANONCODE;
+    assert(o);
+
+    assert(o->op_type < OP_max);
+    assert(o->op_type || o->op_ppaddr == PL_ppaddr[0]);
+
+    assert(o->op_ppaddr);
+    assert(o->op_ppaddr == PL_ppaddr[o->op_type]
+          || o->op_type == 158);
+
+}
+
 OP *
 Perl_ck_anoncode(pTHX_ OP *o)
 {



Ive trimmed the rest of the patch, but it looks basically like this, 
repeated


OP *
Perl_ck_anoncode(pTHX_ OP *o)
{
    PERL_ARGS_ASSERT_CK_ANONCODE;
+   Perl_ck_opany(aTHX_ o);

• rfc - Perl_ck_op_sanity by Jim Cromie