- Following two lines added
DEPENDS+= p5-JSON-MaybeXS-[0-9]*:../../converters/p5-JSON-MaybeXS
DEPENDS+= p5-Term-UI-[0-9]*:../../devel/p5-Term-UI
(upstream)
- Update 5.013 to 5.031
---------------------
5.031 2015-01-08 22:04:30-05:00 America/New_York
- correct a test to avoid testing symlinks on Win32
5.030 2015-01-04 22:31:38-05:00 America/New_York
- fixed [GatherDir]'s handling of symlinks to directories
- [AutoPrereqs] now filters out all namespaces found in contained
modules, not just the one corresponding to the module filename
5.029 2014-12-14 14:44:44-05:00 America/New_York
- fix new error in [PkgVersion] when a module had no package
statements
- further rip out use of JSON.pm
5.028 2014-12-12 19:06:23-05:00 America/New_York
- fix regression in [PkgVersion] that made false-positive
identifications for pre-existing assignments to $VERSION
- try avoid cases in which plugin code directly modifies file list
- switch, tentatively, to JSON::MaybeXS
5.027 2014-12-09 09:30:30-05:00 America/New_York
- fix regression in Plugin->plugin_from_config which started passing a
list of pairs rather than a hashref
5.026 2014-12-08 21:33:55-05:00 America/New_York
- eliminate use of Moose::Autobox
- various small performance optimizations
- add "use_our" option to PkgVersion
5.025 2014-11-10 21:12:14-05:00 America/New_York
- fix file.t failures with perl v5.14 and v5.16's Carp
5.024 2014-11-05 23:08:07-05:00 America/New_York
- add the %Mint stash for minting defaults
- quiet down some low-priority log lines
- teeny tiny optimization by building dist prereqs structure lazily
- avoid ever requiring v0 of ExtUtils::MakeMaker
- fix a module-loading ordering issue in `dzil setup`
5.023 2014-10-30 22:56:42-04:00 America/New_York
- optimizations to loading of heavyweight libraries in cmd line app
- some tests are now skipped on Win32 to avoid filename insanity
- files' added_by data should be more informative
- conflicts with installed code is now detected and/or advertised
5.022 2014-10-27 22:55:53-04:00 America/New_York
- several optimizations to how PPI is used
- handle an empty ABSTRACT better
- now properly merging distmeta fragments together without loss, using
new CPAN::Meta::Merge
- create Makefile.PL and Build.PL files earlier, so they're in the file
list "the whole time"
5.021 2014-10-20 22:43:52-04:00 America/New_York
- improve authordeps' ability to cope with version requirements and
non-default plugin names
- a few improvements to help given by "dzil help COMMAND"
- fixes a situation where exclusion-regexp-building in GatherDir
could mangle the given regexps
- now properly merging distmeta fragments together without loss, using
new CPAN::Meta::Merge (Karen Etheridge)
- [PkgVersion] now properly skips over $VERSION assignments in
comments (Karen Etheridge, github #322)
- the building of manpages is supressed in [MakeMaker]-driven builds
- lazily load quite a few more modules
- avoid using user's ~/.dzil even more
- while building dists for testing, don't bother building man pages
- try harder to notice minimum required perl version
- try harder to delete temporarily directory at the end of testing
- don't treat $VERSION assignments in comments as $VERSION assignments
- listdeps now takes --omit-core to skip core modules
- don't try to use terminal encoding on locale-free systems
- suggest the use of PPI::XS
- speed up and debug behavior of GatherDir
5.020 2014-07-28 20:56:25-04:00 America/New_York
- the default required version for ExtUtils::MakeMaker in [MakeMaker]
has been removed
- load DateTime lazily
- the default required version for Module::Build in [ModuleBuild] has
been lowered
5.019 2014-05-20 21:11:47-04:00 America/New_York
- remove a very brief-lived attempt to double-decode
5.018 2014-05-20 21:07:04-04:00 America/New_York
- attempt to return abstract-from-file as a string, rather than
bytes, which can lead to weirdness (github issue #303)
5.017 2014-05-17 08:35:33-04:00 America/New_York
- dotfiles and dot-directories are now included in sharedirs
- ModuleBuild and MakeMake should not re-build if it isn't needed
- authordeps now better understands "perl" dep
- munging of README is delayed to prevent unneeded work and
complication
- MANIFEST is now treated as a binary file
- 'dzil setup' now warns that credentials are stored in the clear
- MakeMaker should include fewer empty and useless hashrefs
- Makefile.old is now pruned as cruft
5.016 2014-05-05 22:27:06-04:00 America/New_York
- hint about [Encoding] plugin in encoding error message (David
Golden)
5.015 2014-03-30 21:55:36-04:00 America/New_York
- make it easier to have multiple PAUSE configs using UploadToCPAN's
pause_cfg_dir option (thanks, David Golden)
5.014 2014-03-16 16:47:07+01:00 Europe/Paris
- Added 'jobs' argument for 'dzil test' for parallel testing (thanks,
David Golden!)
- add default_jobs attribute to TestRunner role
- fix the behavior of 'dzil add' with more than one file
(thanks, Leon Timmermans!)
Do it for all packages that
* mention perl, or
* have a directory name starting with p5-*, or
* depend on a package starting with p5-
like last time, for 5.18, where this didn't lead to complaints.
Let me know if you have any this time.
Update DEPENDS
Upstream changes:
5.013 2014-02-08 17:08:16-05:00 America/New_York
- META.json is now a UTF-8 file, rather than ASCII
- document the use of filefinders in [PkgVersion], and remove
filtering out of .t, .pod files; do skip non-text files, though
- always load modules in "extra tests" like pod-coverage.t
- PruneCruft also prunes ./fatlib
- avoid being tricked by statements in __END__ section when looking for
variable assignments
- if "dzil install" fails due to exception, it is now propagated
- provide a better error when terminal encoding can't be determined
5.012 2014-01-15 09:58:00-05:00 America/New_York
- when handling a multi-line abstract, fold the lines on whitespace;
previously, the newlines had been left in, which caused downstream
warnings
5.011 2014-01-12 16:09:29-05:00 America/New_York
- ->VERSION is again defined in the tester forms of Builder and Minter
- remove a small obsolete code path from PkgVersion
5.010 2014-01-11 22:06:04-05:00 America/New_York
- stop sharing a reference to cached PPI docs, which led to spooky
action at a distance
- PkgVersion no longer surrounds the new $VERSION assignment with a
bare block
- if there's a blank line after the package statement (and any number
of comment-only lines), PkgVersion will use that for a $VERSION
assignment, rather than insert a new line; this can be made mandatory
with die_on_line_insertion
5.009 2014-01-07 20:21:17-05:00 America/New_York
- include time offset by default in NextRelease
- always pass PPI octets, not text
5.008 2013-12-27 21:57:02 America/New_York
- fix utterly broken `dzil run`
5.007 2013-12-27 20:50:45-05:00 America/New_York
- add the ability to say "dzil run --no-build" to run a command without
building inside the dist dir
(in other words, no `perl Makefile.PL && make`)
- Archive::Tar::Wrapper added as a recommended prereq
- fix :ShareFiles (thanks, Christopher J. Madsen and Karen Etheridge)
- new :AllFiles and :NoFiles filefinders (thanks, Karen Etheridge)
- most files generated by dzil plugins now self-identify with comments
5.006 2013-11-06 09:21:12 America/New_York
- add ->is_bytes to files; shortcut for ->encoding eq 'bytes'
(thanks, David Golden)
- AutoPrereqs will not try scanning binary files (thanks, David Golden)
5.005 2013-11-02 16:32:04 America/New_York
- add --keep-build-dir to "dzil test" and "dzil install"
5.004 2013-11-02 09:59:18 America/New_York
[THIS RELEASE MIGHT BREAK YOUR BUILD]
- stable release of all the v5 changes below; READ THEM!
- by default, NextRelease now adds a trial release marker on trial
releases
- dzil setup will not echo password during setup
5.003 2013-10-30 08:02:59 America/New_York
[THIS RELEASE MIGHT BREAK YOUR BUILD]
- add "dzil --version" support (thanks, Upasana Shukla)
- fix boneheaded mistake that broke listdeps in 5.002 (thanks, Karen
Etheridge)
5.002 2013-10-29 10:35:54 America/New_York
[THIS RELEASE MIGHT BREAK YOUR BUILD]
- perform encoding steps during listdeps
5.001 2013-10-23 17:40:09 America/New_York
[THIS RELEASE MIGHT BREAK YOUR BUILD]
- typo fixes (thanks, David Steinbrunner)
5.000 2013-10-20 08:10:02 America/New_York
[THIS RELEASE MIGHT BREAK YOUR BUILD]
- all files now have content, encoded_content, and encoding attributes
- the Encoding plugin and EncodingProvider role have been added to
allow you to set the encoding on files; the default is UTF-8
- config.ini is assumed to be in UTF-8
- Data::Section sections are assumed to be UTF-8
- the Term chrome should encode input and output
4.300039 2013-09-20 06:05:10 Asia/Tokyo
- tweak metafile generator to keep CPAN::Meta validator happy (thanks,
David Golden)
4.300038 2013-09-08 09:18:34 America/New_York
- add horrible hack to avoid generating non-UTF-8 META.yml; seriously,
don't look at the code! Thanks, David Golden, whose code was simple
and probably much, much saner, but didn't cover as many cases as rjbs
wanted to cover.
4.300037 2013-08-28 21:43:36 America/New_York
- update repo and bugtacker URLs
4.300036 2013-08-25 21:41:21 America/New_York
- read CPAN::Uploader config with CPAN::Uploader, to work with new
trial releases supporting encrypted credential (thanks, Mike Doherty)
- improve tester tests (thanks, Dave O'Neill!)
- use Class::Load instead of Class::MOP
- better error messages when a bundle can't be loaded by @Filter
- make dynamic_config distmeta sticky; once one plugin sets it, it
stays stuck
- add a die_on_existing_version option to PkgVersion
- switch (for now?) "dzil install" to use cpanm
- PkgVersion won't rewrite file contents if nothing was changed
(thanks, Mike Doherty!)
a) refer 'perl' in their Makefile, or
b) have a directory name of p5-*, or
c) have any dependency on any p5-* package
Like last time, where this caused no complaints.
CPAN into devel/p5-Dist-Zilla.
Dist::Zilla builds distributions of code to be uploaded to the CPAN.
In this respect, it is like ExtUtils::MakeMaker, Module::Build, or
Module::Install. Unlike those tools, however, it is not also a system for
installing code that has been downloaded from the CPAN. Since it's only
run by authors, and is meant to be run on a repository checkout rather
than on published, released code, it can do much more than those tools,
and is free to make much more ludicrous demands in terms of prerequisites.
