On Thu, Oct 05, 2006 at 05:41:06AM -0700, Jerry Hedden wrote: > Jerry D. Hedden proposed: > > I'm considering a one line addition to the threads module > > to > overload the string value of a thread to return its > > ID: > > '""' => sub \&tid > > Nicholas Clark remarked: > > I do not wish the *default* behaviour of any part of a > > maintenance release to be incompatible (with the exception > > of bug fixes). > > Jonathan Stowe remarked: > > I guess that most people wouldn't object to the change > > going into blead and essentially the CPAN version can do > > what it wants within reason, but the requirement to not > > change the default behaviours of stuff in maint is always > > going to hold you back > > I think I get it now. The issue isn't about adding the > feature. It just that when added, it shouldn't get > propagated into maint. That should be easy. I'll make a > patch for blead, and the maint pumpking can opt not to use > it. Correct?
Well, that then creates another problem. What $VERSION should threads::shared in maint report? Doing this makes a code fork. I've been adopting the policy of not merging code to released maint unless it *is* an official release of the module on CPAN (for all dual life modules). Whereas there is no code fork if the code is identical, but the *default* behaviour in 5.8.x is not to do this, with the module having an import time option to turn it on. It's trivial for code that knows that it wants this feature from threads::shared to be written to request that threads::shared enable it. Nicholas Clark
