develooper Front page | perl.perl5.porters | Postings from June 2018

Re: [perl #133270] Blead Breaks CPAN: HMBRAND/DBD-CSV-0.53.tgz

Thread Previous | Thread Next
From:
Dave Mitchell
Date:
June 10, 2018 20:36
Subject:
Re: [perl #133270] Blead Breaks CPAN: HMBRAND/DBD-CSV-0.53.tgz
Message ID:
20180610203622.GK26559@iabyn.com
On Sat, Jun 09, 2018 at 11:05:23PM +0100, Dave Mitchell wrote:
> I've now seen a failure under 5.26.0 (one failure in about 200 runs of
> t/82_free_unref_scalar.t).

I've now bisected it (kinda). From at least 5.26.0 until the commit below,
t/82_free_unref_scalar.t would typically trigger a failure once every 100
runs or so, and that failure would typically involve a single double-freed
SV. From the commit through to bleed, around 1 in 4 test runs fails,
typically with several double-freed SVs.

I don't know yet whether blead has introduced or new bug, or whether it
just makes the test more sensitive to an existing bug - a bug which could
be in perl or one or more CPAN modules. :-(


commit e84e4286916d8a219c8a63468807b41df9cde7fe
Author:     David Mitchell <davem@iabyn.com>
AuthorDate: Mon Jul 17 16:33:38 2017 +0100
Commit:     David Mitchell <davem@iabyn.com>
CommitDate: Thu Jul 27 11:30:24 2017 +0100

    Give OP_RV2HV a targ
    
    OP_RV2AV already has one; its not clear why OP_RV2HV didn't.
    Having one means that in scalar context it can return an int value
    without having to create a mortal. Ditto when its doing 'keys %h' via
    OPpRV2HV_ISKEYS.


-- 
All wight. I will give you one more chance. This time, I want to hear
no Wubens. No Weginalds. No Wudolf the wed-nosed weindeers.
    -- Life of Brian

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About