develooper Front page | perl.perl5.porters | Postings from October 2017

[perl #132252] Blead Breaks CPAN: Role::Tiny, Variable::Magic, Moo

Thread Previous
Father Chrysostomos via RT
October 29, 2017 18:40
[perl #132252] Blead Breaks CPAN: Role::Tiny, Variable::Magic, Moo
Message ID:
On Thu, 26 Oct 2017 08:06:54 -0700, wrote:
> On 10/18/2017 02:13 PM, James E Keenan wrote:
> > On 10/18/2017 05:43 AM, Sawyer X wrote:
> >>
> >>
> >> On 10/17/2017 03:55 PM, James E Keenan via RT wrote:
> >>> On Mon, 16 Oct 2017 16:16:18 GMT, wrote:
> >>>> [...]
> >>>> Father C., if you can share your patch, others could help with the
> >>>> patching. (I volunteer.)

All the modules listed in this ticket now have patches:

Haarg says in that he has prepared patches for Moo and Role::Tiny.







> >>> Since this code had no FAILs within the CPAN test suite -- only
> >>> downstream on CPAN.  I think we need to look at the code on CPAN
> >>> which broke to write tests which we can then include in the Perl 5
> >>> test suite to head off such failures in the future.
> >>
> >> I like this idea.

I this case it doesn’t really apply.  In fact, most of the modules are already buggy, because they don’t take into account that, since 5.6 or so, stash elements have contained things other than typeglobs.  In some instances I wrote new tests for the modules that got them to fail with existing perl versions.  The fix to work with the optimisation in these cases was the same as the fix to work with constants in earlier perl versions.

> >>
> >
> > Sawyer,
> >
> > I think it will have to be your call to either (a) do the above, which
> > entails reverting merge commit
> > 1369fd508410a5ab354672cedce158f1e9c653c9 from blead in its entirety
> > (after copying HEAD to a branch); or (b) fixing only a small portion
> > on that commit, as indicated in Father C's most recent posts:
> >
> >
> >
> >
> > (a) has the disadvantage that it puts us into the murkiness described
> > in file:///usr/share/doc/git-doc/howto/revert-a-faulty-merge.html. 
> > (b) has the disadvantage that we have to trust the committer that all
> > *other* parts of 1369fd5 are correct and that we only need to repair
> > one line. I think it is unfortunate that we have been placed in this
> > situation.
> I had asked to temporarily revert it.

I reverted to the old behaviour in commit 6eed25e25 and have now pushed a new sprout/cv-in-stash branch, based on current blead (6e8135a4380966), with the revert reverted.

At this point, how do we proceed?


Father Chrysostomos

via perlbug:  queue: perl5 status: open

Thread Previous Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About