perl@34559

[Nicholas Clark]

I've made a snapshot. Hopefully the final snapshot

http://www.ccl4.org/~nick/P/perl-34559.tar.bz2

Six things that I know of stand between this and RC1

6: Will there be a PathTools release incorporating the Neutrino fix?

5: Will there be a File::Path release incorporating the Neutrino fix?

4: Will anyone on perl5-porters give feedback to Vincent about the 3 ways
to fix http://rt.perl.org/rt3/Ticket/Display.html?id=60034

3: Slaven informs me that change 30539 breaks ex::lib::zip. I don't know
why yet. (Too busy pointing fingers at myself). Possibly SelfLoader.

2: make distclean leaves a couple of directories around

1: perl589delta

Everything else works. Or you didn't tell me, so you get to fix it.

Nicholas Clark

[Nicholas Clark]

On Sat, Oct 25, 2008 at 02:55:40PM +0200, Slaven Rezic wrote:
> Nicholas Clark <ni...@ccl4.org> writes:
>
> > I've made a snapshot. Hopefully the final snapshot
> >
> > http://www.ccl4.org/~nick/P/perl-34559.tar.bz2
>

> Business-ISBN-2.03 seems to be broken by a recent change to 5.8.x,
> 5.10.x and 5.11.x. See
>
> http://bbbike.radzeit.de/~slaven/cpantestersmatrix.cgi?dist=Business-ISBN%202.03;maxver=;reports=1;os=linux;sort=action
>
> (It's not only Andreas' machine, I see the failure on my machine, too)

Well, it worked for me in the UK. Maybe it's only machines in Berlin that fail?
So I tried it on a machine in Vienna, and it failed too. Strange. Maybe there's
something we don't know.

But your report has LANG as en_US.utf8, and even when I set things to
en_GB.utf8 it doesn't fail. However, when I upgrade the passing machine from
Business::ISBN::Data from 1.17 to 20081020 it replicates your fail.
(And I can replicate the fail on 5.8.8)

So I invoke the S.E.P. field again.

Thanks for your continued testing. 5.8.9 will be better for it (even if it's
making me a bit tired currently)

Nicholas Clark

[Nicholas Clark]

On Sun, Oct 26, 2008 at 11:44:56AM +0100, Slaven Rezic wrote:
> Nicholas Clark <ni...@ccl4.org> writes:
>

> > I've made a snapshot. Hopefully the final snapshot
> >
> > http://www.ccl4.org/~nick/P/perl-34559.tar.bz2
> >
>

> IO-Mark-v0.0.1 worked with 5.8.8@RELEASE and fails with 5.8.8@34559.
> binsect says that
>
> p4raw-id: //depot/maint-5.8/perl@30269
>
> is the first bad change. Yes, I know, too many patches in this change.
> Unfortunately the problem does not arise in bleadperl.

I can't replicate it. What failure output does it give, and what's your
perl -V?

Nicholas Clark

[Nicholas Clark]

On Sun, Oct 26, 2008 at 12:39:58PM +0100, Slaven Rezic wrote:
> Nicholas Clark <ni...@ccl4.org> writes:
>
> > On Sun, Oct 26, 2008 at 11:44:56AM +0100, Slaven Rezic wrote:
> > > Nicholas Clark <ni...@ccl4.org> writes:
> > >
> > > > I've made a snapshot. Hopefully the final snapshot
> > > >
> > > > http://www.ccl4.org/~nick/P/perl-34559.tar.bz2
> > > >
> > >
> > > IO-Mark-v0.0.1 worked with 5.8.8@RELEASE and fails with 5.8.8@34559.
> > > binsect says that
> > >
> > > p4raw-id: //depot/maint-5.8/perl@30269
> > >
> > > is the first bad change. Yes, I know, too many patches in this change.

Well, it was the merge of everything in one because all the bugs were fixed.
(Well, that was the plan)

> > > Unfortunately the problem does not arise in bleadperl.
> >
> > I can't replicate it. What failure output does it give, and what's your
> > perl -V?
> >
>

> You can see the failure in all four FAIL reports on:
>
> http://bbbike.radzeit.de/~slaven/cpantestersmatrix.cgi?dist=IO-Mark;maxver=;reports=1;perl=5.8.8
>
> (perl -V output included there)

Ah yes. I can replicate this. But also, I can replicate it in blead on FreeBSD.

Could you reconfigure your blead bisector to compile blead with -Dusefaststdio
and see if you can nail down which change in blead first caused it?

-Dusefaststdio is the default on 5.8.x, but not on 5.10 and onwards.
I think that the choice it makes is only relevant if d_faststdio is defined,
which isn't the case on Linux.

Nicholas Clark