Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 479064 - dev-lang/perl 5.18.1: version bump
Summary: dev-lang/perl 5.18.1: version bump
Status: RESOLVED OBSOLETE
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: [OLD] Development (show other bugs)
Hardware: All Linux
: Normal normal (vote)
Assignee: Gentoo Perl team
URL: https://metacpan.org/release/RJBS/per...
Whiteboard:
Keywords:
Depends on:
Blocks: 481416 483956
  Show dependency tree
 
Reported: 2013-07-31 04:37 UTC by Kent Fredric (IRC: kent\n) (RETIRED)
Modified: 2014-02-25 07:44 UTC (History)
17 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-07-31 04:37:37 UTC
5.18.1 is not out yet, but is expected to hit CPAN in the next week or two.

So this is a pre-emptive bug to start indexing the things that need to change between 5.16 and 5.18.1, of which, most can be based on the difference between 5.16 and the presently-available 5.18.

Reason: I'm not suggesting we waste effort on 5.18.0 with 5.18.1 being so close, especially given how much effort is required to do a major bump. 

Off the top of my head, the things that are relevant to us as downstream are at least:

1. Lots of modules are de-coring/de-dual lifing, and although this release still contains many of them, they now emit loud warnings when loaded, so for these cases the virtuals referring to them should depend *ONLY* on perl-core/* , so that users are installing only the warning-free version of those modules.

2. ~~ is now "experimental", because upstream are looking to change the API details it uses, which means code presently using ~~ may be broken in a future perl realase.

3. Hash Randomization: Perl's Hash implementation has been hugely overhauled, so that the order returned by  keys() and values() is no longer constant, or even constant within the same process, and simply modifying the number of keys in a hash can change the implicit order of the hash, so any naive code that relies on implict ordering will fail. 

Usually, this means you just have to call "sort" more.
Comment 1 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-07-31 04:49:08 UTC
As per: https://metacpan.org/module/RJBS/perl-5.18.0/pod/perldelta.pod#Module-removals 

Virtuals that will need to be modified to adapt to changes in 5.18

Also, any modules using any of the below will now *NEED* to use the virtuals, and can no longer rely on the assumption that it is provided by perl. 


(!) Archive::Extract - virtual/perl-Archive-Extract
(!) B::Lint - virtual/perl-B-Lint
(!) B::Lint::Debug - virtual/perl-B-Lint
(!) CPANPLUS and all included CPANPLUS::* modules - virtual/perl-CPANPLUS
Devel::InnerPackage - virtual/perl-Module-Pluggable
(!) Log::Message - virtual/perl-Log-Message
(!) Log::Message::Config - virtual/perl-Log-Message
(!) Log::Message::Handlers - virtual/perl-Log-Message
(!) Log::Message::Item - virtual/perl-Log-Message
(!) Log::Message::Simple - virtual/perl-Log-Message-Simple
Module::Pluggable - virtual/perl-Module-Pluggable
Module::Pluggable::Object - virtual/perl-Module-Pluggable
(!) Object::Accessor - virtual/perl-Object-Accessor
(!) Pod::LaTeX - virtual/perl-Pod-LaTeX
(!) Term::UI - virtual/perl-Term-UI
(!) Term::UI::History - virtual/perl-Term-UI

(!)  = Presently only available on ::perl-experimental
Comment 2 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-07-31 05:25:09 UTC
Virtuals that need to be updated to map to new providers:

https://metacpan.org/module/RJBS/perl-5.18.0/pod/perldelta.pod#Updated-Modules-and-Pragmata

(+) virtual/perl-Archive-Tar-1.900.0 
(!) virtual/perl-B-Debug-1.180.0
(!) virtual/perl-Carp-1.290.0
(+) virtual/perl-CGI-3.630.0
(+) virtual/perl-Compress-Raw-Bzip2-2.60.0
(+) virtual/perl-Compress-Raw-Zlib-2.60.0
(+) virtual/perl-CPAN-Meta-Requirements-2.112.0
(*) virtual/perl-Data-Dumper-2.145
(+) virtual/perl-DB_File-1.827.0
(+) virtual/perl-Digest-MD5-2.520.0
(*) virtual/perl-Digest-SHA-5.840.0
(+) virtual/perl-Encode-2.490.0
(*) virtual/perl-ExtUtils-CBuilder-0.280.210
(+) virtual/perl-ExtUtils-ParseXS-3.180.0
(+) virtual/perl-File-Fetch-0.380.0
(*) virtual/perl-File-Temp-0.230.0
(!) virtual/perl-HTTP-Tiny-0.25.0
(*) virtual/perl-IO-1.280.0
(+) virtual/perl-IPC-Cmd-0.800.0
(+) virtual/perl-Memoize-1.30.0
(+) virtual/perl-Module-Build-0.400.300
(*) virtual/perl-Module-CoreList-2.900.0
(+) virtual/perl-Module-Load-Conditional-0.540.0
(+) virtual/perl-Module-Metadata-1.0.11
(*) virtual/perl-net-ping-2.410.0
(*) virtual/perl-Pod-Simple-3.280.0
(+) virtual/perl-Safe-2.350.0
(+) virtual/perl-Scalar-List-Utils-1.270.0
(*) virtual/perl-Socket-2.9.0
(*) virtual/perl-Storable-2.410.0
(+) virtual/perl-Sys-Syslog-0.320.0
(+) virtual/perl-Term-ANSIColor-4.20.0
(+) virtual/perl-Test-Harness-3.260.0
(+) virtual/perl-Text-Tabs+Wrap-2012.81.800
(+) virtual/perl-Time-Local-1.230.0


(!) = ::perl-experimental and needs perl-5.18.0 added
(*) = this version is not in tree yet
(+) = this version is in tree, and needs perl-5.18.0 added
Comment 3 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-07-31 05:29:13 UTC
Additionally, virtual/perl-Version-Requirements is no longer shipped by dev-lang/perl, and is deprecated upstream in favour of CPAN-Meta-Requirements.

so:

virtual/perl-Version-Requirements should depend only on perl-core/Version-Requirements after 5.18, and we should be looking to tree cleaning it (eventually).
Comment 4 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-08-13 10:32:13 UTC
5.18.1 is out, hacking on ::perl-experimental getting virtuals up-to-date
Comment 5 Patrick Lauer gentoo-dev 2013-08-14 06:21:55 UTC
ebuilds now in tree and masked together with perl-5.18:

=virtual/perl-Attribute-Handlers-0.940.0
=virtual/perl-CPAN-2.0.0
=virtual/perl-Devel-PPPort-3.200.0-r2
=virtual/perl-Dumpvalue-1.170.0-r1
=virtual/perl-ExtUtils-Constant-0.230.0-r3
=virtual/perl-File-Temp-0.230.0
=virtual/perl-Filter-Simple-0.890.0
=virtual/perl-IO-Zlib-1.100.0-r3
=virtual/perl-Locale-Maketext-Simple-0.210.0-r3
=virtual/perl-MIME-Base64-3.130.0-r3
=virtual/perl-Math-BigInt-FastCalc-0.300.0-r2
=virtual/perl-Math-Complex-1.590.0-r2
=virtual/perl-Module-Loaded-0.80.0-r2
=virtual/perl-Package-Constants-0.20.0-r3
=virtual/perl-Perl-OSType-1.3.0
=virtual/perl-Pod-Escapes-1.40.0-r3
=virtual/perl-Test-Simple-0.980.0-r3
=virtual/perl-Text-Balanced-2.20.0-r3
=virtual/perl-Text-ParseWords-3.280.0-r1
=virtual/perl-Tie-RefHash-1.390.0-r1
=virtual/perl-Time-HiRes-1.972.500-r2
=virtual/perl-digest-base-1.170.0-r2
=virtual/perl-if-0.60.200
=virtual/perl-libnet-1.220.0-r2
=virtual/perl-parent-0.225.0-r4
=virtual/perl-threads-1.860.0-r2
=virtual/perl-ExtUtils-CBuilder-0.280.210
=virtual/perl-IO-1.280.0
=virtual/perl-Module-CoreList-2.900.0
Comment 6 Patrick Lauer gentoo-dev 2013-09-05 07:16:36 UTC
5.18.1 has been unmasked. I did lots of checking to see if it's sane, upgraded two chroots, as far as I'm concerned there are no big issues.

Keeping this bug open for now, let's see what issues our users find  :)
Comment 7 Mikle Kolyada (RETIRED) archtester Gentoo Infrastructure gentoo-dev Security 2013-09-08 18:35:12 UTC
I mask perl-5.18 again. It needs more testing and we should add miss modules.
Comment 8 Laurent Bachelier 2013-09-09 04:22:58 UTC
I am getting, running stable:

!!! The following installed packages are masked:
- virtual/perl-File-Temp-0.220.0-r2::gentoo (masked by: package.mask)
/usr/portage/profiles/package.mask:
# Mikle Kolyada <zlogene@gentoo.org> (08 Sep 2013)
# Untested / unconfirmed major bump of perl #479064
# Needs proper testing before being unleashed


It's required by many packages.
Comment 9 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-09-09 04:30:19 UTC
This really shouldn't be unmasked till we at least have *SOME* of the patches from 5.16 ported to 5.18.

Otherwise, people are just going to be bumping into bugs we've already got solutions for, and already got bugs for in bugzilla, ... simply because we haven't forward-ported the patches.
Comment 10 Kent Fredric (IRC: kent\n) (RETIRED) gentoo-dev 2013-09-09 04:33:17 UTC
( And some of those are subtle things you wont simply see with basic testing, such as failure to respect *FLAGS and failure to create the right .so files, etc , etc  )
Comment 11 Jason A. Donenfeld gentoo-dev 2013-09-09 10:29:21 UTC
Well this is a massive pain for me. I've already upgraded _several machines_, some of which are very slow so compilations take a long long time (embedded mips systems), before the packages were remasked.

Now they're masked, creating big emerge problems. The only feasible solution for me is to just unmask everything myself.

But now there is talk of subtle errors. When you get around to fixing those errors, would you mind rev-bumping each and every package that you modify, so that I get those fixes during the usual update process?

Thanks.
Comment 12 Franz Trischberger 2013-09-11 06:41:03 UTC
(In reply to Kent Fredric from comment #10)
> ( And some of those are subtle things you wont simply see with basic
> testing, such as failure to respect *FLAGS and failure to create the right
> .so files, etc , etc  )

And I already wondered, why claws-mail could not read the addressbook index, disabled plugins, lost subfolders of one imap inbox (not on the server, just locally) and with that the setting for the SENT MAIL folder. And all after perl update + perl-cleaner + rebuild of claws-mail (removal of old python version + PYTHON_TARGETS + me using --newuse). I already suspected disc failure, but everything else is consistent.
Comment 13 Mikle Kolyada (RETIRED) archtester Gentoo Infrastructure gentoo-dev Security 2014-02-25 07:44:32 UTC
bug 502380