[BUG] branch_cs opcode is buggy and unused

[cotto]

The branch_cs opcode causes a nice segfault when called with a random string. When I looked into it a little more deeply, it became apparent that this opcode isn't used at all during make test. If it's worth fixing and testing then we should do that. Otherwise it needs to be deprecated.

[whiteknight]

Can we get some kind of answer about this one before the release? Do we want to deprecate branch_cs, or do we want to fix/test it? If the former, I would like to get in the deprecation notice before tuesday.

[coke]

FYI, deprecation notices must be present in stable releases before something can be removed. Next stable release is 1.4; next monthly release is 1.2.

[whiteknight]

Okay, got the answer from Allison today in #ps. We're removing this opcode. I'll put in the deprecation notice ASAP.

[coke]

from src/runcore/main.c:

 /* if we have fallen out with resume and we were running CGOTO, set
         * the stacktop again to a sane value, so that restarting the runloop
         * is ok. */
        if (interp->resume_flag & RESUME_RESTART) {
            if ((int)interp->resume_offset < 0)
                Parrot_ex_throw_from_c_args(interp, NULL, 1,
                    "branch_cs: illegal resume offset");
            stop_prederef(interp);
        }

branch_ch is just a string here, but does this mean this whole conditional can be removed? If not, can someone suggest a better error message?

[whiteknight]

patch to remove branch_cs opcode

[whiteknight]

I've attached a patch that should perform the removal of this opcode (except for the error message that Coke pointed out above).

DEPRECATED.pod says that this is eligible in 1.5, although I think I added that notice so I could have been wrong. Can I remove this now, or do we need to wait till after the August release?

[coke]

This can be removed now. Go for it.

[cotto]

branch_cs was removed in r40432.