Monday 1st August 2011
Local Packages
Updated perl-Module-Load to 0.20:
Fix CPAN RT#69886: trailing '::' silently ignored
Rebuild for various clean-ups such as removal of macros, fixing dist tag for CentOS 6, using patch instead of iconv:
unrar
weblint
weblint++
xv
xxdiff
Tuesday 2nd August 2011
Local Packages
Updated my builder to use kernel 2.6.40 so that I can use file capabilities on tmpfs (Bug #648653) and hence build packages for Fedora 15 onwards more quickly
Updated my builder configuration for Fedora 3, 4, 5 and 6 to exclude libnet10 from Fedora Extras as it can get pulled in by yum in mock when a package has a build requirement of libnet-devel, even though there's higher versioned package of libnet-devel available (Bug #229297)
Updated my repository-building script to remove empty garbageid directories left around by createrepo (Bug #581628)
Updated mgdiff to use the tarball from ftp.x.org rather than the one I originally pulled from a SuSE source RPM a few years ago
Updated perl-Test-NoWarnings to 1.03:
Test::Builder and Test::Tester dependencies were the wrong way around (build vs. runtime)
I also reinstated the build requirements perl(Pod::Simple), perl(Test::CPAN::Meta), perl(Test::MinimumVersion) and perl(Test::Pod) that had been dropped for the 1.02 build, and run the release tests during the build
- Rebuild for various clean-ups such as removal of macros, fixing dist tag for CentOS 6:
kernel-dummy
shorten
symlinks
tcptraceroute
torrentsniff
tzip
Wednesday 3rd August 2011
Fedora Project
Updated perl-Compress-Raw-Lzma to 2.036 for F-15, built against the forthcoming xz 5.0.3 update for that release
Updated perl-Module-Metadata to 1.000005 in F-16 and Rawhide:
Localize $package::VERSION during version discovery
Fix references to Module::Build::ModuleInfo (CPAN RT#66133)
Added 'new_from_handle()' method (CPAN RT#68875)
Improved documentation (SYNOPSIS, broke out class/object method, and other minor edits)
I also changed the package to install into the vendor directories rather than the perl directories, as per just about every other module in Fedora
Local Packages
New package perl-Data-Section-Simple (0.02)
New package perl-Module-Extract-VERSION (1.01)
New package perl-Test-Version (1.0.0)
Updated curl to add a new option CURLOPT_GSSAPI_DELEGATION (Bug #719939)
Updated dkms to 2.2.0.2
Updated perl-Coro to 6.04:
Change how Coro patches AnyEvent condvars for compatibility with AnyEvent 6.x
Update ecb.h to no longer include <pthread.h> in case WinNT.h hasn't been included
Updated perl-CPAN-Meta to 2.112150:
Stringify any objects encountered during conversion (CPAN RT#67295)
Fixed some broken URLs in documentation and for LGPL in legacy conversion (CPAN RT#68738)
Fixed invalid private Pod in CPAN::Meta::Validator (CPAN RT#65925)
Clarified that file paths in the 'provides' section must be in Unix-style (i.e. forward slashes) (CPAN RT#69045)
Replaced examples using Module::Build::ModuleInfo with Module::Metadata (CPAN RT#66135)
I added new test build requirements: perl(Pod::Coverage::TrustPod), perl(Test::CPAN::Meta), perl(Test::Pod::Coverage), perl(Test::Portability::Files) and perl(Test::Version) ≥ 0.04
Updated perl-Locale-US to 2.112150 (largely rewritten, fixing CPAN RT#56989); I added new build requirements perl(Data::Section::Simple) and perl(Test::More), packaged the LICENSE file and added a patch to support building with ExtUtils::MakeMaker < 6.30
Updated perl-Module-Metadata to 1.000005 as per the Fedora version
Updated perl-Net-FTPSSL to 0.18:
- More fixes/clarifications to the POD text
Fixed minor wildcard issue with LIST and NLST if "+" was in the filename
Added LocalAddr option to new(); works against both the command channel and the data channel
Added restart() function to be compatible with Net::FTP for restarts
Added ability to restart downloads/get() where the previous attempt left off
Added ability to restart uploads/put() where the previous attempt left off
Added append() command [APPE]; it also uses OFFSET, but doesn't send the REST command to the server
Masks out the user name used when generating a log file, similar to how this is already done for the password except that it must also hide it in the response to USER and PASS; it's only masked when written to the log - message() and last_message() still return the user name unmasked in any response string!
Fixed a bug in t/10-complex.t and added optional environment variables for some of the connection values to make it easier to quickly re-run tests during development without having to key in answers to questions with no good default values over and over again
Added SSL_Client_Certificate to support client certificates on the data channel as well as the command channel (CPAN RT#69327)
Deprecated SSL_Advanced in favour of SSL_Client_Certificate; SSL_Advanced is now just an alias for SSL_Client_Certificate for backwards compatibility; the new name better describes what we are trying to do with it now that this feature is supported
Wrote t/20-certificate.t to test this out, though this and t/10-complex.t are probably incompatible in most settings - anyone else trying to use this test case will have to modify the code to point to their certificate and provide its password, since the upstream test client certificate isn't going to be published
Calls to command() now return a reference to this class so that you can code something like:
if ($ftps->command("NOOP")->response() == CMD_OK)Modifed Net::FTPSSL to use the new command()->response() format internally
Thursday 4th August 2011
Fedora Project
Submitted package review requests for perl-Data-Section-Simple, perl-Module-Extract-VERSION and perl-Test-Version
Local Packages
New package perl-Devel-Hide (0.0008)
Updated perl-List-MoreUtils to 0.33:
Updated can_xs to fix a bug in it
Updated perl-namespace-clean to 0.21, dropping the no-longer-needed patch for building with old versions of ExtUtils::MakeMaker, adding build requirement perl(Devel::Hide), needed for the new pure-perl test, and dropping no-longer-needed build requirements perl(Pod::Coverage::TrustPod), perl(Test::EOL), perl(Test::NoTabs), perl(Test::Pod) and perl(Test::Pod::Coverage):
Only invoke the deleted sub stashing if we run under a debugger (avoid runtime penalty of Sub::Name/Sub::Identify)
Spelling fixes (CPAN RT#54388)
When B::Hooks::EndOfScope is not available, switch to a simple tie() of %^H
Document how to disable the tie() fallback
- Rebuild for various clean-ups such as removal of macros, fixing dist tag for CentOS 6:
pyserial
python-crypto
python-pyasn1
Updated svnmailer to fix "ImportError: cannot import name exceptions" traceback using patch based on one attached to Gentoo Bug #209857
Friday 5th August 2011
Local Packages
Updated perl-Coro to 6.05:
Blush: condvar values would not be propagated from send to recv any more
Use exponential increase for the readline buffer length in Coro::Handle; also reduce initial allocation to 1020 from 4096 bytes
Updated perl-ExtUtils-ParseXS to 3.02, except for RHL-9 and RHEL-3 since 3.x now require perl ≥ 5.8.1:
Major refactoring of the whole code base; it finally runs under 'use strict' for the first time!
Typemaps can now be embedded into XS code using a here-doc like syntax and the new "TYPEMAP:" XS keyword
Move typemap handling code to ExtUtils::Typemaps with full object-oriented goodness
Check API compatibility when loading XS modules using the XS_APIVERSION_BOOTCHECK macro if on a new-enough perl
Fixed compiler warnings in XS
- Spell-check
Define PERL_UNUSED_ARG for pre-3.8.9 perls, which should fix the tests
Add a check to Makefile.PL that scans @INC to determine whether the new xsubpp will be shadowed by another, existing xsubpp and warn the user vehemently
Add some diagnostics when xsubpp fails to load a current-enough version of ExtUtils::ParseXS
Patches from CPAN RT#53938, CPAN RT#61908, attempts to fix Win32 problems
Require perl 5.8.1
I excluded /usr/bin/xsubpp and its manpage from the package for the time being as they would conflict with the main perl or perl-devel packages (Bug #728393)
Updated perl-HTML-SimpleLinkExtor to 1.25:
Fix clear_links so it actually clears the links
Convert most tests to use subtest and Test::More 0.96
Clean-ups for the move to Github
I added a patch to support building with Test::More versions as old as 0.47 on RHL-9; that's the oldest I tried
Saturday 6th August 2011
Local Packages
Updated perl-BerkeleyDB to 0.49:
- Documentation updated courtesy of Mike Caron
Croak if attempt to freeze BerkeleyDB object (CPAN RT#69985)
- Rebuild for various clean-ups such as removal of macros, fixing dist tag for CentOS 6:
libnet
libxml2
libxslt (discovered rpm-build Bug #728959 whilst doing this one, resulting in some unexpectedly-packaged documentation)
nmap
perl-YAML-LibYAML
Sunday 7th August 2011
Fedora Project
Mario Blättermann kindly reviewed my perl-Data-Section-Simple package and approved it, and I built it for F-14, F-15, F-16, Rawhide, EPEL-4, EPEL-5 and EPEL-6
Local Packages
New package perl-Module-Manifest (1.08)
New package perl-Pod-Wordlist-hanekomu (1.110090)
New package perl-Test-DistManifest (1.011)
Updated perl-Text-CSV_XS to 0.83a:
Fix spurious auto_diag warning (CPAN RT#69673)
Tested with 50 versions of perl, including 1.15.1
Monday 8th August 2011
Fedora Project
Updated gtkwave in F-16, Rawhide and EPEL-6 to 3.3.24:
- Improve the searching for the TCL libraries (when using stubs)
Fixed bug where Tcl_GetString was substituted with brace removal preprocessing when unnecessary (would break addSignalsFromList, etc.)
Became owner of perl-Tree-DAG_Node and perl-Test-Warn in EPEL-4 (as Spot no longer has any interest in EL-4) and built them - needed for perl-Module-Manifest
Became owner of perl-Module-Manifest in EPEL, and built it for EPEL-4, EPEL-5 and EPEL-6
Submitted a review request for perl-Test-Vars, which was promptly approved by Emmanuel Seyman and I built it for F-14, F-15, F-16, Rawhide and EPEL-6 (no builds for EPEL-4 or EPEL-5 as it requires perl ≥ 5.10.0)
Local Packages
New package perl-Test-Vars (0.001)
Updated gtkwave as per the Fedora version
Updated perl-Params-Check to 0.32:
Removed unnecessary use of Data::Dumper
Updated perl-Test-Spelling to add a patch for releases with hunspell < 1.2.8, in which hunspell doesn't treat the "C" locale as being equivalent to the "en_US" locale, causing test failures on Fedora 9 and 10 because it can't find a working speller for the default locale
Rebuild perl-Test-Requires without a workaround previously added for the Test::Spelling issue above
Tuesday 9th August 2011
Fedora Project
Updated perl-Package-Stash-XS to 0.23 in F-16 and Rawhide:
- Fix the test for scalar values, again
- Disallow assigning globrefs to scalar glob slots (this doesn't actually make any sense)
Updated perl-Package-Stash to 0.31 in F-16 and Rawhide:
Fix ->add_symbol('$foo', qr/sdlfk/) on 5.12+
Fix ->add_symbol('$foo', \v1.2.3) on 5.10+
Local Packages
New package perl-Test-Has-Version (0.012)
Updated perl-Coro to 6.06:
Cygwin unfortunately patches the stack at runtime, so we use the pthreads backend, which is an order of magnitude slower; unfortunately, Cygwin's pthread implementation isn't very complete either, so allocate the stack twice just to be sure
Updated perl-Package-Stash-XS to 0.23 as per the Fedora version
Updated perl-Package-Stash to 0.31 as per the Fedora version
Updated perl-Pod-Wordlist-hanekomu to BR: perl(Test::HasVersion) and perl(Test::Vars) for full release test coverage
Wednesday 10th August 2011
Fedora Project
Became co-maintainer of the new branches I requested for perl-Test-DistManifest and built it for F-14, F-15, EPEL-4, EPEL-5 and EPEL-6
Became co-maintainer of the new EPEL branches I requested for perl-Test-HasVersion and built it for EPEL-4, EPEL-5 and EPEL-6
Submitted a review request for perl-Pod-Wordlist-hanekomu
Submitted a review request for perl-Test-CPAN-Meta-JSON
Local Packages
New package perl-Test-CPAN-Meta-JSON (0.10)
Thursday 11th August 2011
Fedora Project
Became owner of perl-JSON in new EPEL-4 branch and built it
Updated perl-Test-Spelling in F-14 and F-15 to 0.14; it's been in F-16 and Rawhide for months now, including through the Perl mass rebuild, and Test::Spelling ≥ 0.12 was needed for perl-Test-Mojibake
Submitted a review request for perl-Test-Mojibake
Updated tidyp to fix mangling of non-ASCII output filenames (Bug #725651) since an upstream release addressing the issue doesn't seem to be on the cards any time soon
Local Packages
New package perl-Test-Mojibake (0.3)
New package perl-Unicode-CheckUTF8 (1.03)
I raised CPAN RT#70210 to ask upstream to specify the license for this module
Updated perl-ExtUtils-ParseXS to 3.03, reinstating %{_bindir}/xsubpp and its manpage if we have perl ≥ 5.12.4, as per F-15 onwards (conflict with perl-devel in Bug #728393 was fixed in perl-5.12.4-160):
Test fix: Try all @INC-derived typemap locations (CPAN RT#70047)
Updated perl-JSON-XS to 2.32:
- Fix a bug in the initial whitespace accumulation
Updated perl-Test-Version to not run the author/release tests when bootstrapping, BR: perl(Test::DistManifest) unconditionally now that it's available locally, and to additional BR's for improved release test coverage:
aspell-en/aspell-en-gb
perl(Pod::Wordlist::hanekomu)
perl(Test::CPAN::Meta::JSON)
perl(Test::Mojibake)
perl(Test::Vars)
Friday 12th August 2011
Fedora Project
Iain Arnell kindly reviewed and approved my perl-Module-Extract-VERSION, perl-Pod-Wordlist-hanekomu, perl-Test-CPAN-Meta-JSON and perl-Test-Mojibake packages
Built perl-Module-Extract-VERSION for EPEL-4, EPEL-5, EPEL-6, F-14, F-15, F-16 and Rawhide
Built perl-Pod-Wordlist-hanekomu for F-16 and Rawhide; builds for other releases are waiting for perl-Test-DistManifest to be in the buildroot (Bug #672543)
Built perl-Test-CPAN-Meta-JSON for EPEL-5, EPEL-6, F-14, F-15, F-16 and Rawhide; the EPEL-4 build is waiting for perl-JSON to be in the buildroot (https://admin.fedoraproject.org/updates/perl-JSON-2.17-1.el4)
Built perl-Test-Mojibake for F-16 and Rawhide; builds for other releases are waiting for perl-Test-DistManifest to be in the buildroot (Bug #672543)
Local Packages
Updated perl-Mail-IMAPClient to 3.29:
CPAN RT#69876: ENVELOPE as part of fetch_hash convenience method: added Mail::IMAPClient::BodyStructure::Envelope->parse_string($str) convenience method for handling ENVELOPE data from fetch_hash
CPAN RT#68310: folders() should not call exists()/STATUS: use selectable() instead of exists() in call
Ensure separator is set properly in folders()
selectable now properly checks for \Noselect flag
Update folders() POD to match implementation behaviour
CPAN RT#68648: CAPABILITY after authenticate
Delete cache after State set toAuthenticate
State() is no longer an auto-generated method
Fix CPAN RT#68755: Provided socket loses blocking in 3.19-3.28
Updated perl-Module-Extract-VERSION to filter bogus provide for perl(ExtUtils::MakeMaker::_version) (Bug #728286)
Updated python-zope-interface to 3.6.5:
LP#804951: InterfaceClass instances were unhashable under Python 3.x
LP#811792: Work around buggy behaviour in some subclasses of zope.interface.interface.InterfaceClass, which invoke __hash__ before initializing __module__ and __name__; the workaround returns a fixed constant hash in such cases, and issues a UserWarning
LP#804832: Under PyPy, zope.interface should not build its C extension; also prevent attempting to build it under Jython
Add a tox.ini for easier xplatform testing
Fix testing deprecation warnings issued when tested under Py3K
- Initially I found that the test suite failed (in a new test just added):
====================================================================== FAIL: test_hash_missing_required_attrs (zope.interface.tests.test_interface.InterfaceTests) ---------------------------------------------------------------------- Traceback (most recent call last): File "/builddir/build/BUILD/zope.interface-3.6.5/src/zope/interface/tests/test_interface.py", line 420, in test_hash_missing_required_attrs self.assertEqual(len(warned), 1) AssertionError: 0 != 1 ---------------------------------------------------------------------- Ran 167 tests in 0.209s FAILED (failures=1)I added a patch to resolve this and sent it upstream (LP#825249)
I also added a second patch to remove the problematic test altogether if we have python 2.4, as it uses the "with" statement introduced in Python 2.5; all other tests pass with 2.4, and upstream skips the test on Python 2.5 for other reasons so it's not that important
Sunday 14th August 2011
Local Packages
Updated perl-AnyEvent to 6.0 but after discussion with upstream it appears that the 6.0 release shouldn't actually have happened so I reverted it
Updated python-zope-interface to 3.7.0:
LP#825249: fix test_hash_missing_required_attrs
Monday 15th August 2011
Local Packages
Created an initial java-1.7.0-oracle spec, based on a java-1.7.0-sun spec sent to me by Knut Jørgen Bjuland; unfortunately the old Java program I use is EMC NetWorker's management console, and that hasn't been ported to Java 7 yet (it doesn't work) so I'm holding off doing a full write-up on it
Rebuilt libgcrypt 1.5.0 due to rpm bug #728707
Tuesday 16th August 2011
Local Packages
New package perl-Perl6-Junction (1.40000)
New package perl-Pod-Coverage-Moose (0.02)
Updated curl to fix SIGSEGV of curl -O -J given more than one URL (Bug #723075), introduce the --delegation option of curl (Bug #730444), and initialize NSS with no database if the selected database is broken (Bug #728562)
Updated libssh2 to 1.2.9:
Added libssh2_session_set_timeout() and libssh2_session_get_timeout() to make blocking calls get a timeout
configure and pkg-config: fix $VERSION
s/\.NF/.nf/ to fix wrong macro name caught by man --warnings
keepalive: add first basic man pages
sftp_write: flush the packetlist on error
sftp_write: clean offsets on error
msvcproj: added libs and debug stuff
SCP: fix incorrect error code
session_startup: init state properly
sftp_write_sliding: send the complete file
userauth_keyboard_interactive: skip code on zero length auth
_libssh2_wait_socket: fix timeouts for poll() uses
agent_list_identities: fix out of scope access
_libssh2_recv(): handle ENOENT error as EAGAIN
userauth_keyboard_interactive: fix buffer overflow
- Removed man pages for non-existing functions!
gettimeofday: fix name space pollution
_libssh2_channel_write: handle window_size == 0 better
Updated perl-Class-Load to 0.08:
The previous version was missing a prereq declaration for Data::OptList (CPAN RT#70285)
The distribution has been converted to Dist::Zilla and a bunch of release tests added, so packaging-wise there were a lot of changes:
This release by DROLSKY -> update source URL
Package new documentation: LICENSE and README
- Add build requirements for new release tests and run them:
perl(Pod::Coverage::Moose)
perl(Test::CPAN::Changes)
perl(Test::EOL)
perl(Test::NoTabs)
perl(Test::Pod)
perl(Test::Pod::Coverage)
perl(Test::Requires)
perl(Test::Spelling) and aspell/aspell-en-gb
Add patch for building with ExtUtils::MakeMaker < 6.30
Add patch for building with Test::More < 0.88
Add patch for building without Pod::Coverage::Moose
Add patch for building without Test::Requires
- Add patch for fixing spell checker word list
Update patch for building without Test::Fatal (using Test::Exception instead)
Wednesday 17th August 2011
Fedora Project
Updated perl-CPAN-Meta-YAML in Rawhide to install to vendor perl directories rather than private perl directories
Updated perl-Perl-OSType in Rawhide to build with perl(Pod::Coverage::TrustPod) unconditionally now that it's available in EPEL-4
Updated perl-Test-Requires in EPEL-6 to build using hunspell-en rather than aspell-en now that RHEL-6's perl-Test-Spelling package has been updated to use hunspell rather than aspell (Bug #731272)
Local Packages
Updated c-ares to 1.7.5:
Detection of semicolon comments in resolv.conf
Avoid using system's inet_net_pton affected by the WLB-2008080064 advisory
Replacement ares_inet_net_pton affected by the WLB-2008080064 advisory
Replacement ares_inet_ntop affected by potential out of bounds write
Added install target to Makefile.msvc
Only fall back to AF_INET searches when looking for AF_UNSPEC addresses
Fixed ares_parse_*_reply memory leaks
Use correct sizeof in ares_getnameinfo()
- IPv6-on-windows: find DNS servers correctly
man pages: docs for the c-ares utility programs
getservbyport replacement for Win CE
config_sortlist: (win32) missing else
advance_tcp_send_queue: avoid NULL pointer dereference
configure: fix a bash-ism
ares_expand_name: Fix encoded length for indirect root
Updated java-1.6.0-sun to Java SE 6 Update 27 along with tzupdater 1.3.40 but unfortunately it doesn't work with my test case of the EMC NetWorker Management Console (Could not initialize class COM.legato.gwt.gtpanel.GtUtils), which worked fine with Update 26, so I'm not updating SunJava6OnFedora until I've identified where the problem is
Updated perl-CPAN-Meta-YAML to install to vendor perl directories rather than private perl directories
Thursday 18th August 2011
Local Packages
Updated perl-Test-Most to 0.25:
Add timeit() function to time individual behaviours
Friday 19th August 2011
Local Packages
Unbundled perl-DateTime-Locale and perl-DateTime-Timezone from perl-DateTime as per the Fedora versions; they're all built from their own SRPMs now
Updated perl-Locale-Maketext-Lexicon to 0.91:
Added support to the TT2 parser for Mojolicious style tags
- Documentation changes
I originally worked on 0.90 but there was a problem with test counts in t/5-extract.t so I sent a patch upstream, and 0.91 was promptly issued with the fix
I added perl(Template) ≥ 2.20 as a build requirement from F-11 onwards (the package is too old in earlier releases) for additional test coverage
Updated perl-MIME-Types to 1.32:
Explain how to use MIME::Types in mod_perl; when you do not read the documentation about mod_perl/fork it will work as always, but inefficiently
Saturday 20th August 2011
Fedora Project
Built perl-Pod-Wordlist-hanekomu (1.110090) and perl-Test-Mojibake (0.3) for F-14 and F-15 now that perl-Test-DistManifest has been pushed to stable for those releases
Sunday 21st August 2011
Fedora Project
Patched curl with an additional upstream fix for SIGSEGV of curl -O -J given more than one URL (Bug #723075); builds for F-15 and F-16 completed successfully but the Rawhide build (same code as F-16) failed its test suite and I'll need to look into that. I've not pushed any updates for this as there's already an important update in the queue and I don't want to obsolete it for this relatively trivial issue
Local Packages
Updated curl as per the Fedora version
Updated perl-Module-CoreList to 2.55:
- Updated for v5.15.2
Updated perl-version to 0.94:
Prevent DIE handlers in user code from tripping up loading version, and add test to ensure that handling is correct (CPAN RT#70260)
Tweak POD and add LICENSE to Makefile.PL (CPAN RT#70120)
Monday 22nd August 2011
Fedora Project
Patched curl in F-14 with an additional upstream fix for SIGSEGV of curl -O -J given more than one URL (Bug #723075) though due to issues with the test suite it took me 11 attempts to get it to build
Looked into updating perl-Spreadsheet-ParseExcel to 0.59 for Bug #731907 and found that it needed Digest::Perl::MD5, which isn't currently packaged in Fedora. It's supposed to replaceable by Digest::MD5, which is in Fedora but unfortunately Spreadsheet::ParseExcel uses private data from within Digest::Perl::MD5 rather than using the proper API so using Digest::MD5 doesn't work. I created a perl-Digest-Perl-MD5 package and Petr Sabata kindly reviewed it.
Tuesday 23rd August 2011
Fedora Project
Built curl in Rawhide at the second attempt today (4 attempts altogether); similar problems to those affecting the F-14 build yesterday
Updated perl-Test-Spelling to 0.15 in F-16 and Rawhide:
Begin adding actual tests (hilariously, adding the suggested t/pod-spell.t to this dist to test itself found a typo: "stopwards")
I added new build requirements of perl(Test::Tester) and an English dictionary
Built new package perl-Digest-Perl-MD5 (1.8) for EPEL-4, EPEL-5, EPEL-6, F-14, F-15, F-16 and Rawhide
Updated perl-Spreadsheet-ParseExcel to 0.59 in EPEL-6, F-16 and Rawhide to address Bug #731907
Local Packages
Updated perl-Test-Spelling to 0.15 as per the Fedora version, and I also needed to add various patches to support dictionaries in older releases that didn't have all the necessary words to pass the t/pod-spell.t test, building with perl < 5.8.3 and building with Test::More < 0.88
Wednesday 24th August 2011
Fedora Project
Submitted a review request for perl-Unicode-CheckUTF8, which makes perl-Test-Mojibake faster if available
Updated spamass-milter in all branches to add a patch that helps with users with a Postfix MTA and users doing SMTP AUTH but not using the milter's -I option to not pass mail from authenticated users through SpamAssassin; Postfix doesn't support Sendmail's {auth_ssf} macro that's used to add the (authenticated bits=nnn) information into the dummy Received: header that the milter passes to SpamAssassin, so in its absence, the milter now just adds (authenticated) rather than omitting the information altogether, which is sufficient for SpamAssassin to consider the sender as a trusted source (Bug #730308)
Local Packages
Updated perl-Moose to 2.0203:
is_class_loaded now properly detects packages that have a version object in their $VERSION
Fix XS compilation under bleadperl
Updated perl-Pod-Simple to 3.19:
Fixed occasional test failure when tests are run in parallel (HARNESS_OPTIONS=j6:c)
Added "pod_handler" option, a callback that can be used to inspect the content on a "=pod" line
- Added tests for proper line-ending parsing from the previous release
Added the "parse_empty_lists" option (disabled by default); when enabled, Pod::Simple will recognize empty lists (that is, blocks with '=over' and '=back' but nothing in between)
Added the "whiteline_handler" attribute, much like "code_handler", "cut_handler", and "pod_handler", except it's triggered on lines that contain only whitespace characters
Added "raw" attribute to L<> treelet that contains the L<>'s raw contents
Pod directives autoclosed by Pod::Simple are now denoted as such by the new "fake_closer" attribute
Fixed incompatibility with Pod::Simple::HTMLBatch in Pod::Simple::XHTML
I had to add a patch to support building with Test::More < 0.88 (CPAN RT#70456)
Updated spamass-milter as per the Fedora version
Thursday 25th August 2011
Local Packages
Updated perl-ExtUtils-ParseXS to 3.04:
- Properly strip trailing semicolons from inputmaps, which could trigger warnings (errors in strict C89 compilers) due to additional semicolons being interpreted as empty statements
Now detects and throws a warning if there is a CODE section using RETVAL, but no OUTPUT section (CPAN RT#69536)
Uses the explicit XS_EXTERNAL macro (from perl 5.15.2 and newer) for XSUBs that need to be exported; defines XS_EXTERNAL to fall back to XS where that is not available
Introduces new EXPORT_XSUB_SYMBOLS XS keyword that forces exported XSUB symbols; it's a no-op on perls before 5.15.2
Try to fix regression for input-typemap override in XS argument list (CPAN RT#70448)
- Explicit versions in submodules to fail early if something goes wrong
Updated perl-XML-LibXML to 1.85, adding a patch to fix the missing symbol SvPVx_nolen on perl < 5.8.8 (raised as CPAN RT#70476):
Gracefully handle returned undef()s in the read callback under -w ($^W); add t/49callbacks_returning_undef.t (CPAN RT#70321)
Add a patch from Mithaldu to get XML::LibXML to compile on Win32 (CPAN RT#70141)
Updated perl-XML-LibXML to 1.86, dropping the now-redundant patch for (CPAN RT#70476):
Changed SvPVx_nolen() to SvPV_nolen() in LibXML.xs for better compatibility; SvPVx_nolen() appears to be undocumented API (CPAN RT#70476)
Friday 26th August 2011
Fedora Project
Built perl-Pod-Wordlist-hanekomu for EPEL-4, EPEL-5 and EPEL-6 now that perl-Test-DistManifest is available there
Built perl-Test-CPAN-Meta-JSON for EPEL-4 now that perl-JSON is available there
Built perl-Test-Mojibake for EPEL-4, EPEL-5 and EPEL-6 now that perl-Test-DistManifest is available there
PPTP Client Project
Updated pptp-release to fix the dist tag for builds with CentOS 6 and Scientific Linux
Updated pptpconfig not to use macros for commands
- Created Fedora 16 repositories
Local Packages
Updated perl-AnyEvent to 6.01:
Incompatible Change: verbosity levels have been redefined; see AnyEvent (PERL_ANYEVENT_VERBOSE) and AnyEvent::Log for details
New function: AnyEvent::log
New module: AnyEvent::Log, for simple logging needs
New env variable, PERL_ANYEVENT_LOG, for fine-grained logging configuration
New env variable, PERL_ANYEVENT_HOSTS, to override the place where the /etc/hosts file can be found
An empty PERL_ANYEVENT_RESOLV_CONF now selects the default configuration
Environment variables can now be specified using AE_xyz instead of PERL_ANYEVENT_xyz, subject to some rules
Lots of new AnyEvent::Debug functionality
AnyEvent::Strict now checks for modifications of $_ by the callback (by setting it to readonly while executing the callback, if possible)
AnyEvent::Socket::resolve_sockaddr and all functions using it now support /etc/hosts overrides
Do not run event loop detection multiple times when programs call their cached copies of AnyEvent::detect
parse_hostport now accepts absolute paths as unix domain sockets
Suppress the idiotic warning inside POE by patching the POE::Kernel run flags instead of asking the user to jump through contortions
Add Task::Weaken as dependency (as recommends only at this time)
Replace PF_UNSPEC by 0, for good old Minix
- This major update also had a bunch of packaging changes:
Drop now-redundant patch removing FLTK2 references
Package manpages for AnyEvent::Impl::FLTK2 and AnyEvent::Log
Add build and runtime dependencies for perl(Task::Weaken)
Make the build dependency on perl(Net::SSLeay) unconditional as although the SSL test still emits a "Too late to run CHECK block" warning prior to perl 5.10, the test still passes
Add build dependencies on perl(Event::Lib), perl(Glib) ≥ 1.210, perl(POE) and perl(Tk) where possible (i.e. pull in from Fedora/EPEL) for the event loop tests
Add a build dependency on perl(Time::HiRes)
Filter the bogus provide of perl(DB)
- These changes were on top of the unpublished changes of 6.0 on August 14th, which for the record were:
Incompatible Change: AnyEvent::DNS resource records now include the TTL as fourth element - this affects ->resolve directly, and ->request and ->dns_unpack indirectly; convenience functions such as AnyEvent::DNS::{a,txt,mx,srv} are not affected, and neither is AnyEvent::Socket
Incompatible Change: backend authors now should not implement one_event or loop, but instead the AnyEvent::CondVar::_wait and _poll methods
Deprecation: please change "use AnyEvent::Impl::Perl" to "use AnyEvent::Loop" and "AnyEvent::Impl::Perl::loop" to "AnyEvent::Loop::run" in your programs, for future compatibility
- Idle watcher emulation was borked
AnyEvent::Strict now checks the AE::xxx functions as well
Querying too-long domain names would barf deep inside AnyEvent::DNS; since this is hard to avoid for applications, AnyEvent:.DNS::request now checks for too-long domain names explicitly and passes an undef to the callback instead
Make AnyEvent::Util more robust against EINTR
New function: AnyEvent::postpone
New env variable, PERL_ANYEVENT_DEBUG_SHELL, to automatically start a debug shell
New env variable, PERL_ANYEVENT_DEBUG_WRAP, to automatically enable watcher instrumentation/wrapping (see AnyEvent::Debug::wrap)
IO::Async sometimes cancels the wrong timer when you call cancel_timer; invest extra overhead to work around this peculiar behaviour
Split the pure perl backend into a loop module and an impl module (AnyEvent::Loop and AnyEvent::Impl::Perl)
Some read types (regex, netstring, json, storable) could erroneously keep an AE::Handle object alive on their own
Add more workarounds around child watcher bugs in IO::Async and (newly-introduced) in POE
Use glib child watchers and try to work around its limitations, if possible
Add support for FLTK (fltk2) via AnyEvent::Impl::FLTK2
Many read types in AnyEvent::Handle would malfunction during transitions to TLS because of a "delete $self->{rbuf}"
Improve condvar blocking wait performance (probably only noticeable with faster backends) and also make it easier for event loops that do not support blocking waits
PERL_ANYEVENT_MODEL now also allows full module names, although the usefulness of this feature is not yet clear
Add tests for all supported event loops, conditional on PERL_ANYEVENT_LOOP_TESTS
Add t/80_ssltext.t
No longer use AUTOLOAD for the initial loading (for no very good reason)
tcp_server now tries to unlink unix domain sockets when it gets destroyed
- Packaging changes for 6.0:
Patch AnyEvent::Impl::FLTK2 out of Makefile.PL as upstream forgot to ship it
Package new manpage for AnyEvent::Loop
Add new build dependencies for tests: perl(EV), perl(Event), perl(Net::SSLeay)
I then updated perl-AnyEvent again, this time to 6.02:
log_to_syslog now accepts facility strings
log_to_syslog would not allow facilities - "$facility|$level" literally means that string
The CPAN parser got confused and thought we provide Sys::Syslog; try to work around it
Rename AnyEvent::Impl::FLTK2 to FLTK
Updated perl-Moose to 2.0204:
Validating duck_type type constraint turned out to work only by accident, and only when not running under the debugger; this has been fixed
Loosen the dependency on ExtUtils::ParseXS
Updated proftpd to fix the dist tag for builds with CentOS 6 and Scientific Linux
Previous Month: July 2011
Next Month: September 2011