Bug#654387: [perl #71172] New lvalue subroutine roadblock in 5.10.1 breaks important functionality

[Father Chrysostomos via RT]

On Sun Jan 08 14:56:04 2012, sprout wrote:
> On Sun Jan 08 14:49:34 2012, dom wrote:
> > On Mon May 30 21:35:27 2011, sprout wrote:
> > > This has hopefully now been fixed by commit b724cc1, which takes away
> > > the TEMP check in pp_leavesublv.
> > > 
> > > Now there is a warning (introduced shortly before in commit 8fe85e3),
> > > which is only triggered if a value is assigned to a returned TEMP.
> > > 
> > > So the error has basically been downgraded.
> > 
> > And this was re-fixed in fd6c41ce5607a262d2124271d289aa2a0213c049 in a 
> > different way.
> > 
> > There has been a request for the earlier patch for 5.10.1 on this
ticket 
> > to be applied in Debian (http://bugs.debian.org/654387) and I would be 
> > interested in opinions about whether this is sensible to apply.
> 
> The earlier patch is no good.  It allows existing bugs (see
> <https://rt.perl.org/rt3/Ticket/Display.html?id=78194>) to occur in more
> cases.

Clarification: If, by the earlier patch, you mean b724cc1, that is no good.

The earlier patch in this ticket
(<https://rt.perl.org/rt3/Ticket/Display.html?id=71172#txn-773706>),
which checks for magic, might be safer than fd6c41ce56.

> 
> The later patch is probably OK for 5.10.1, but I have not checked
> whether it depends on any other commits.  I don’t think it does.
> 


-- 

Father Chrysostomos


---
via perlbug:  queue: perl5 status: resolved
https://rt.perl.org:443/rt3/Ticket/Display.html?id=71172