On Wed, Sep 17, 2008 at 4:00 PM, Daniel J. Luke <[EMAIL PROTECTED]> wrote: > On Sep 17, 2008, at 3:58 AM, Vincent Lefevre wrote: >> On 2008-09-15 12:25:37 -0400, Daniel J. Luke wrote: >>> Except that you say it as a blanket statement even though you've been >>> shown at least one case where the port requires the newer version (svk). >> No, upstream doesn't claim any such dependency on a newer version of >> a Perl module. You're just seeing a bug, that needs to be fixed. > The Makefile.PL that ships with svk clearly states a requirement for > File::Temp 0.17, which is why the p5-file-temp-svkonly port was created (as > 0.17 is newer than what was distributed with perl 5.8.8).
I'd find it reasonable though, to agree on a policy which forbids ports overwriting the contents of their dependents voluntarily. And if it is really desirable somewhere, then it should be marked clearly as a workaround. Your example is quite good in that respect. But ports like p5-test-harness make the missunderstanding quite easy that you have to install them, when you need the perl module Test::Harness. The name is simply missleading. It is rather a "monkey patch" of the original perl installation. So we should maybe: 1) outrule ports overwriting dependents 2) make a exception for special cases, but those require an additional name suffix like -patch to make the indended use clear. 3) if 1 or 2 are violated this is a bug and has to be fixed Bug #12710 lists quite a few of these ports, (only perl ones): http://trac.macports.org/ticket/12710 What do you think of the proposal, e.g. as an extension to the guide? -- Florian Ebeling [EMAIL PROTECTED] _______________________________________________ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
