dynoplibs in core or no?

Jonathan Leto jaleto at gmail.com
Wed Jun 2 01:42:06 UTC 2010


On Tue, Jun 1, 2010 at 4:22 PM, Will Coleda <will at coleda.com> wrote:
> My vote is to resurrect the getstd* opcodes.
> Since the general consensus seems to be "doesn't matter", let's move
> these back.

Not to be a pest, but the "open" opcode becoming a dynop is what broke
PL/Parrot recently. There is nothing wrong with open being a dynop per
se, but a long standing bug with dynops that rarely got tickled is now
much more important. If that bug was fixed, "open" being a dynop would
be all gravy.

I think it would be nice if basic I/O opcodes became "plain ops"
again, until the bug with dynops in PBC's is fixed. I think other HLL
authors will run into the same problem that PL/Parrot is having, and
currently it looks like fixing dynops in PBC's is non-trivial and may
require some deprecations.

I am currently attempting to write some tests for the dynop bug after
talking with chromatic++ about what the simplest test case would look


Jonathan "Duke" Leto
jonathan at leto.net

More information about the parrot-dev mailing list