Dropping 5.5.x

[Nicholas Clark]

On Wed, Mar 11, 2009 at 02:40:08PM -0400, David Golden wrote:

> Back on-side the argument -- if a major security bug were found going
> all the way back to 5.6.2, would there be a 5.6.3?  If so, then I
> would argue not to drop support for 5.6.  If not -- if the answer is
> "tough, upgrade to 5.8.10 or 5.10.1 for the fix" then I think that 5.6
> could be considered end-of-life.

No, likely there would not be. If "major security bug" means something
with a CVE, then historically what has happened is a patch against each
affected released version, fixing just that issue. Not a full blown release.
Even for a current stable branch.

A 5.6.3 is more likely to be released to cope with cumulative toolchain
"fixage" that breaks the core perl build. Such as GNU removing + options
from whatever-it-was, or new an exciting output from gcc's dependency
checker. Certainly no bug fixes.

There are good arguments for keeping the suite of CPAN modules in a really
stable stable release branch just the same, as one use people have for such
releases is testing their code on a the modern OS they have locally, where
they know it needs to run on an existing constrained install on an older
system.

ie this is an argument against worrying about whether the current version
of Test::Harness or any other CPAN module runs on 5.6.anything or
5.005_anything, because all that really matters in those situations is
that the core will build at all, to run the modules it shipped with back then.

Nicholas Clark