On Fri, Sep 10, 2010 at 06:07:34AM +0200, Florian Ragwitz wrote: > Jesse Vincent <jesse@fsck.com> writes: > > > On Fri, Sep 10, 2010 at 10:52:39AM +1000, Tony Cook wrote: > >> Do we remove -Dmad or fix it? > > > > I would be deeply, deeply sad if mad were removed. > > Why? MAD, even with the bitrot it's suffered because its tests have a non-core dependency, is still our best hope for source-to-source translation for Perl. It's capable of giving us full ASTs for the actual source developers write and turning those ASTs back into Perl source code. Gerard's experiments around using it to "port" Perl 5 code to Kurila programmatically offer a pretty compelling vision of a way to deal with future Perl 5 language changes. With MAD, it should be(come) possible to dump arbitrary source code with "5.10 MAD" and pull it back in with "5.14 MAD" and _correctly_ update existing code to match the evolving language. And yes, if we had tools that could upgrade old code to the current version of the language, that would go a very, very long way toward solving my backward-compatibility issues. --Thread Previous | Thread Next