Front page | perl.perl5.porters |
Postings from January 2005
On Jan 27, 2005, at 3:56 AM, Jos I. Boumans wrote: > > On Jan 26, 2005, at 9:00 PM, Ken Williams wrote: > >> On Jan 25, 2005, at 3:26 PM, John Peacock wrote: >> >>> Jos I. Boumans wrote: >>>> I suppose if Module::Build::Base would say, at the top: >>>> $VERSION = $Module::Build::VERSION >>>> this would make EU::MM, M::IV, M::L::C and of course CPANPLUS::inc >>>> find a 'real' version declaration, rather than something evil that >>>> will blow away UNIVERSAL::VERSION. >>> >>> Yes, that would be an appropriate fix to M::B::Base. >> >> That won't work. All those tools evaluate the version line in >> isolation, not by loading the modules and checking the value of >> $VERSION. In other words, when Module::Build::Base's version is >> checked, Module::Build won't be loaded, so the version would be >> undef. > > Well, the above one line fix adresses 2 issues, and they both 'do > work': > 1) Under normal operation, M::B::Base is loaded from M::B, so > the $VERSION variable would be defined. Anyone using M::B > programatically > would find what is expected. Right. > 2) All tools using the EU::MM 'standard' regex would find a valid > version > declaration. Except that it won't have the right value, it'll be undef. Tools like the META.yml builders, search.cpan.org, etc. that report versions of modules would all be fooled. I think the only sane fix here is to make sure that whatever code, be it in M::B or CP+ or perl, is evaluating these lines, doesn't cause a segfault when it makes a mistake. Bad version lines are always going to be out there, even if we legislate against them. As for M::B, of course I can change it, and I suppose I probably should. Split that code over two lines or something. -KenThread Previous | Thread Next