Re: File::Copy::Recursive broken on FreeBSD and other OSes; working on replacement

[email protected] (Karen Etheridge) Fri, 20 Apr 2018 14:30:05 +0200
Newsgroups perl.cpan.workers
Message-ID <CAPJsHfDAucm60p0LjjGbS7tK7cEqu8TL_FiFqxDVd=boumcWFg@mail.gmail.com>
--0000000000001e51b6056a46d9fd
Content-Type: text/plain; charset="UTF-8"

A new File::Copy::Recursive has been released -- has the threat now passed?

On Wed, Apr 18, 2018 at 7:25 PM, James E Keenan <[email protected]> wrote:

> This message is intended for people working on the Perl toolchain and, in
> particular, for attendees at the Perl Toolchain Summit starting tomorrow in
> Oslo.
>
> The test-against-dev work I have been conducting for several months has
> identified at least one CPAN distro, high up in the toolchain, which is
> broken on non-Linux operating systems.  The reasons for this breakage are
> sub-optimal code in modules or test suites, not "Blead Breaks CPAN."  Such
> breakage prevents the distro from being installed and means that its
> revdeps are never reached during the following cases:
>
> * Installation via 'cpan' shell:  Distro fails and is graded as FAIL and
> reported to metabase.  Assuming on 'force install' is attempted on distro,
> revdep is graded as DISCARD and no report is sent to metabase or to
> revdep's maintainers.
>
> * Installation via 'cpanm':  Distro fails and is graded as FAIL.  Revdep
> is not attempted and is not graded anywhere.  You have to examine build.log
> for messages reporting "Bailing out ...."  No report is sent to metabase or
> to maintainers.
>
> * Because test-against-dev relies on 'cpanm', if a distro in, say, the top
> 1000 of the CPAN river fails, its revdeps are never reached, meaning their
> code does not get exercised against the latest monthly development release.
>
> The CPAN distro of greatest concern right now is File-Copy-Recursive(
> http://search.cpan.org/~dmuey/File-Copy-Recursive-0.40/), currently
> ranked about #128 in the CPAN river.  On FreeBSD (and perhaps other OSes),
> FCR's current version (0.40 as of Apr 17 2018) has an unfixed bug (
> https://rt.cpan.org/Ticket/Display.html?id=123964) leading to test
> failures and failure to install.  Unless a user chooses to '--force'
> install FCR, that prevents its revdeps from being installed in an automated
> manner or tested against dev/blead.
>
> Over 6000 CPAN distributions have direct or indirect dependencies on FCR,
> including CPAN-Reporter, Test-File-ShareDir, DateTime, Dist-Zilla and
> Catalyst-Devel.  So thousands of CPAN distros cannot be installed on
> FreeBSD and other systems.
>
> The author of FCR is aware of the bug ticket and Tom Hukins provided a
> patch which I have tested successfully on FreeBSD.  But the patch has not
> been applied as of yesterday.  As we are now in the countdown to
> perl-5.28.0, we really need to be able to assess the impact of release
> candidates on CPAN -- and not just on Linux.
>
> Earlier this week I wanted to resume sending CPANtesters reports from
> FreeBSD-11.1.  I tried to install CPAN-Reporter but could not, due to the
> bug in its dependency, FCR.  To address that I have uploaded to CPAN a new
> CPAN distro, File-Copy-Recursive-Reduced (FCR2), which provides
> stripped-down versions of 2 FCR subroutines, fcopy() and dircopy(), which
> are used in the CPAN-Reporter test suite.  I then branched CPAN-Reporter to
> create a version which uses the FCR2 versions of these functions rather
> than FCR.  That makes CPAN-Reporter installable again. I have therefore
> filed:
>
> https://github.com/cpan-testers/CPAN-Reporter/issues/88
> and
> https://github.com/cpan-testers/CPAN-Reporter/pull/90 (a better p.r. than
> my earlier /89)
>
> I hope that during the PTS you can review and apply this p.r. and upload a
> new version of CPAN-Reporter.  This will once again permit automated
> installation on FreeBSD and other non-Linux operating systems.  It will
> also help us to assess perl-5.28's possible impact on that distro and its
> revdeps.
>
> Thank you very much.
> Jim Keenan
>

--0000000000001e51b6056a46d9fd
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_default" style=3D"font-size:small">A n=
ew File::Copy::Recursive has been released -- has the threat now passed?<br=
></div></div><div class=3D"gmail_extra"><br><div class=3D"gmail_quote">On W=
ed, Apr 18, 2018 at 7:25 PM, James E Keenan <span dir=3D"ltr">&lt;<a href=
=3D"mailto:[email protected]" target=3D"_blank">[email protected]</a>&gt;</=
span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8e=
x;border-left:1px #ccc solid;padding-left:1ex">This message is intended for=
 people working on the Perl toolchain and, in particular, for attendees at =
the Perl Toolchain Summit starting tomorrow in Oslo.<br>
<br>
The test-against-dev work I have been conducting for several months has ide=
ntified at least one CPAN distro, high up in the toolchain, which is broken=
 on non-Linux operating systems.=C2=A0 The reasons for this breakage are su=
b-optimal code in modules or test suites, not &quot;Blead Breaks CPAN.&quot=
;=C2=A0 Such breakage prevents the distro from being installed and means th=
at its revdeps are never reached during the following cases:<br>
<br>
* Installation via &#39;cpan&#39; shell:=C2=A0 Distro fails and is graded a=
s FAIL and reported to metabase.=C2=A0 Assuming on &#39;force install&#39; =
is attempted on distro, revdep is graded as DISCARD and no report is sent t=
o metabase or to revdep&#39;s maintainers.<br>
<br>
* Installation via &#39;cpanm&#39;:=C2=A0 Distro fails and is graded as FAI=
L.=C2=A0 Revdep is not attempted and is not graded anywhere.=C2=A0 You have=
 to examine build.log for messages reporting &quot;Bailing out ....&quot;=
=C2=A0 No report is sent to metabase or to maintainers.<br>
<br>
* Because test-against-dev relies on &#39;cpanm&#39;, if a distro in, say, =
the top 1000 of the CPAN river fails, its revdeps are never reached, meanin=
g their code does not get exercised against the latest monthly development =
release.<br>
<br>
The CPAN distro of greatest concern right now is File-Copy-Recursive(<a hre=
f=3D"http://search.cpan.org/~dmuey/File-Copy-Recursive-0.40/" rel=3D"norefe=
rrer" target=3D"_blank">http://sea<wbr>rch.cpan.org/~dmuey/File-Copy-<wbr>R=
ecursive-0.40/</a>), currently ranked about #128 in the CPAN river.=C2=A0 O=
n FreeBSD (and perhaps other OSes), FCR&#39;s current version (0.40 as of A=
pr 17 2018) has an unfixed bug (<a href=3D"https://rt.cpan.org/Ticket/Displ=
ay.html?id=3D123964" rel=3D"noreferrer" target=3D"_blank">https://rt.cpan.o=
rg/Ticket/Di<wbr>splay.html?id=3D123964</a>) leading to test failures and f=
ailure to install.=C2=A0 Unless a user chooses to &#39;--force&#39; install=
 FCR, that prevents its revdeps from being installed in an automated manner=
 or tested against dev/blead.<br>
<br>
Over 6000 CPAN distributions have direct or indirect dependencies on FCR, i=
ncluding CPAN-Reporter, Test-File-ShareDir, DateTime, Dist-Zilla and Cataly=
st-Devel.=C2=A0 So thousands of CPAN distros cannot be installed on FreeBSD=
 and other systems.<br>
<br>
The author of FCR is aware of the bug ticket and Tom Hukins provided a patc=
h which I have tested successfully on FreeBSD.=C2=A0 But the patch has not =
been applied as of yesterday.=C2=A0 As we are now in the countdown to perl-=
5.28.0, we really need to be able to assess the impact of release candidate=
s on CPAN -- and not just on Linux.<br>
<br>
Earlier this week I wanted to resume sending CPANtesters reports from FreeB=
SD-11.1.=C2=A0 I tried to install CPAN-Reporter but could not, due to the b=
ug in its dependency, FCR.=C2=A0 To address that I have uploaded to CPAN a =
new CPAN distro, File-Copy-Recursive-Reduced (FCR2), which provides strippe=
d-down versions of 2 FCR subroutines, fcopy() and dircopy(), which are used=
 in the CPAN-Reporter test suite.=C2=A0 I then branched CPAN-Reporter to cr=
eate a version which uses the FCR2 versions of these functions rather than =
FCR.=C2=A0 That makes CPAN-Reporter installable again. I have therefore fil=
ed:<br>
<br>
<a href=3D"https://github.com/cpan-testers/CPAN-Reporter/issues/88" rel=3D"=
noreferrer" target=3D"_blank">https://github.com/cpan-tester<wbr>s/CPAN-Rep=
orter/issues/88</a><br>
and<br>
<a href=3D"https://github.com/cpan-testers/CPAN-Reporter/pull/90" rel=3D"no=
referrer" target=3D"_blank">https://github.com/cpan-tester<wbr>s/CPAN-Repor=
ter/pull/90</a> (a better p.r. than my earlier /89)<br>
<br>
I hope that during the PTS you can review and apply this p.r. and upload a =
new version of CPAN-Reporter.=C2=A0 This will once again permit automated i=
nstallation on FreeBSD and other non-Linux operating systems.=C2=A0 It will=
 also help us to assess perl-5.28&#39;s possible impact on that distro and =
its revdeps.<br>
<br>
Thank you very much.<br>
Jim Keenan<br>
</blockquote></div><br></div>

--0000000000001e51b6056a46d9fd--