Migrating to Perl 5.20/5.22

[David Evans]

On 7/14/15 7:32 AM, Daniel J. Luke wrote:
>> On Jul 14, 2015, at 5:37 AM, Mojca Miklavec <mojca at macports.org> wrote:
>> We should start migrating other ports that depend on Perl to a newer
>> version of Perl, but the question is: should we go to 5.22 or 5.20?
>> David proposed to go to 5.20 because 5.22 might be too new and not so
>> well tested yet.
> 5.22
>
> 5.22 is upstream’s ‘current’ version, and so by our normal policy it’s what we should pick. The only reason to go with 5.20 instead was if there was a known problem with 5.22 that we couldn’t patch (and even then, we should probably just address it with upstream and go with 5.22).
>
> Once we move to 5.22, we should get rid of all of the old perls and just switch perl5 to be 5.22 and stop trying to support multiple perl versions.
>
> --
> Daniel J. Luke
> +========================================================+
> | *---------------- dluke at geeklair.net ----------------* |
> | *-------------- http://www.geeklair.net -------------* |
> +========================================================+
> |   Opinions expressed are mine and do not necessarily   |
> |          reflect the opinions of my employer.          |
> +========================================================+
>
>
>
> _______________________________________________
> macports-dev mailing list
> macports-dev at lists.macosforge.org
> https://lists.macosforge.org/mailman/listinfo/macports-dev
I agree with you in principle but ...

My point in being cautious has to do with fact that perl5.22 has just 
recently been released as the latest stable version.  I have confidence 
in perl5.22
itself but the fact that it is stable says nothing about whether the 
various Perl modules work with it properly.  They are completely 
untested on Macports
with respect to this version other than to say that they build. Because 
of the apathy towards supporting Perl modules, in general, by the 
MacPorts maintainer community (952/1373 modules are nomaintainer), the 
only time modules actually get tested is when they are updated.  Maybe.

And this testing is really cursory.

I am concerned that switching to perl5.22 immediately will inevitably 
cause unexpected consequences in terms of module useability.

My compromise proposal is to switch to perl5.20 as the default now and 
drop all Perl versions except perl5.20 and perl5.22.  Then plan to drop 
perl5.20
in say three to six months, after there is time for upstream module 
maintainers to address remaining perl5.22 issues, and leave perl5.22 as 
the sole Perl version supported.  After that, update the sole Perl port 
not at the time of the next stable period, but, again, 3 to 6 months 
thereafter for the same reasons.

An ongoing program of testing existing modules against the current 
unstable Perl version would be a further improvement in quality 
control/assurance but I doubt we have the interest or manpower available 
to do it.

Enough discussion.  Let's make a decision and get on with it.

Dave