On Tue, May 07, 2002 at 09:13:28PM +0100, Tim Bunce wrote: > On Tue, May 07, 2002 at 05:08:08PM +0100, Nicholas Clark wrote: > > The dog doesn't read p5p, so I think I can get away with it. Well, he did get 1 walk this evening. (I left my FreeBSD box building 5.6.1 with threads) > Ah, I missed that (too fast with the delete key when skimming my mailboxes). > > + } else if (type >= SX_ERROR && cxt->ver_minor > STORABLE_BIN_MINOR) { > + if (cxt->accept_future_minor < 0) > + cxt->accept_future_minor > + = (SvTRUE(perl_get_sv("Storable::accept_future_minor", > + TRUE)) > + ? 1 : 0); > + if (cxt->accept_future_minor == 1) { > + CROAK(("Storable binary image v%d.%d contains data of type %d. " > + "This Storable is v%d.%d and can only handle data types up to %d", > + cxt->ver_major, cxt->ver_minor, type, > + STORABLE_BIN_MAJOR, STORABLE_BIN_MINOR, SX_ERROR - 1)); > + } > + } > > I've not looked at the code in context, but if type >= SX_ERROR > then don't you *have* to croak? I think the accept_future_minor > test only applies at the point the cxt->ver_minor is set from the > header. Yes, you're right, you do *have* to croak. The conditional code is deciding whether to croak with a the new message you see, or the traditional "Corrupt Storable binary" message. I didn't think that the "corrupt" message was correct (let alone helpful) if the problem was a new data type. > > What version number should it be released as? > > I'd vote for 2.00 since it's a significant improvement in many ways > from the 1.01 that's 'out there'. It'll also align it with the > value of STORABLE_BIN_MAJOR (which hopefully now won't change for > a long time). I'm feeling conservative - at the moment the .pm thinks it's 1.015 Nicholas Clark -- Even better than the real thing: http://nms-cgi.sourceforge.net/Thread Previous | Thread Next