Upstream changes:
1.05 Sun Nov 15 12:54 2009
- Minor developer doc tweaks.
- Minor XS refactoring
1.04_05 Mon Nov 9 20:10 2009
- Fixes for perls < 5.10:
=> No entersub optimization
=> Do no use precalculated hashes
- Updated entersub optimization
- Remove brain-damaged double-hashing
- Minor portability fixlets
1.04_04 Thu Nov 5 18:00 2009
- Fixes for non-threaded perls
(no need for locks, perl_mutex not even defined).
1.04_03 Tue Nov 3 22:32 2009
** This release features some very radical changes. Test well. **
- Replace use of perl hashes in the global hash key name storage with
a full-blown, separate implementation of a hash table
(Steffen, Chocolateboy)
- Similarly, throw out the SV's for simple C strings.
- Add a global lock for all modifications to global data structures:
- The above three items fix RT #50454 (serious threading issues).
- Add support for alternate use Class::XSAccessor { ... } syntax
(Adam K)
1.04_02 Mon Sep 7 11:35 2009
** This release features some very radical changes. Test well. **
- Significant optimization by replacing the relevant entersub ops
with stripped down versions (Chocolateboy)
1.04_01 Mon Sep 7 11:35 2009
** This release features some very radical changes. Test well. **
- More aggressive OPTIMIZE flags if possible (Chocolateboy)
- Added shorthand syntax for getters, setters, accessors, and predicates
where the attribute has the same name as the method (Chocolateboy)
- Remove dependency on AutoXS::Header.
- Merge Class::XSAccessor::Array into this distribution.
- Refactored the XS to remove duplicate code.
- Refactored the perl code in XSAccessor.pm and Array.pm to remove
duplicate code (=> Heavy.pm).
- Upgrade Devel::PPPort/ppport.h (Chocolateboy)
1.04 Thu Jun 11 16:40 2009
- Fix a bunch of warnings thanks to a heads up from
Marcela Maslanova.
LibMicro is a portable set of microbenchmarks that many Solaris engineers
used during Solaris 10 development to measure the performance of various
system and library calls. LibMicro was developed by Bart Smaalders and
Phil Harman as part of their "If Linux is faster it's a Solaris bug
performance" campaign.
Upstream changes:
=====================================
2009-10-14T04:25:48.467603Z ptl-1_030
=====================================
2009-10-14 04:24:18 (r2744) by rcaputo
New version for a new release.
2009-10-13 12:43:42 (r2743) by bingosnet
Explicitly bind to loopback address for the concurrent TCP tests
2009-10-11 18:53:43 (r2741) by rcaputo
Fix Marco's name. Oops!
2009-10-07 01:50:42 (r2736) by rcaputo
Set version for dev release.
2009-10-06 03:50:33 (r2735) by rcaputo
Make the tests generally ASSERT_DEFAULT clean. POE_TRACE_FILENAME hid
a multitude of sins in the test output file. Update the vim options
lines while I'm in there.
2009-10-06 03:39:18 (r2733) by rcaputo
Redirect STDERR while testing some errors. ASSERT_DEFAULT is on and
making noise when things are called wrong, even when we do it on
purpose.
2009-10-05 05:32:30 (r2727) by rcaputo
Add an option to enable all the broken MSWin32 tests.
Set POE_DANTIC=1 before running "make test" to watch the pretty
fireworks. Oooh! Ahhhhhh!
2009-10-02 03:47:17 (r2707) by rcaputo
[rt.cpan.org 48715] Apply Craig's patch to skip the IPv6 tests if
IPv6 isn't actually enabled. The port autodetect code to didn't
handle failure due to IPv6 unavailability.
2009-09-29 15:13:53 (r2703) by rcaputo
!!! Adjust tests for mark-and-sweep session GC changes.
Some of the tests are sensitive to session GC timing.
2009-09-16 15:11:30 (r2691) by rcaputo
Test that POE::Kernel->run() without sessions returns right away.
=====================================
2009-09-07T19:22:59.577723Z ptl-1_022
=====================================
2009-09-07 19:13:43 (r2677) by rcaputo
Bump up the version, and prevent perl-reversion from seeing test
version strings (that are commented out, no less).
2009-09-06 03:32:57 (r2675) by rcaputo
[rt 36648] Add tests for the new Client and Server TCP component
capabilities.
2009-09-05 17:56:52 (r2672) by rcaputo
[rt 49384] Explicitly bind to 127.0.0.1 in tests, reported
anonymously. [cpan testers 5241203] Bind to INPORT_ANY to avoid
"address already in use" failures, reported by Oliver Paukstadt.
2009-09-04 07:49:18 (r2671) by rcaputo
Make write-combining test optional. The timing of some event loops
triggers write-combining, and others don't. Resolves rt.cpan.org
28804 (finally!), reported by Martijn van Beers.
2009-09-04 04:28:55 (r2668) by rcaputo
Tk and fork() don't get along well. Skip some fork() based tests when
Tk is loaded, to avoid XIO errors that have shown up in FreeBSD.
2009-09-03 16:08:06 (r2665) by rcaputo
Skip pty tests in z_leolo_wheel_run.pm on MSWin32. Change some skip
messages.
2009-08-08 08:25:22 (r2625) by rcaputo
Skip a known issue regarding pseudy-ttys, Event, and Mac OS X. On OS
X, poll() returns POLLNVAL for pseudo-ttys. Event cancels its
watchers in this case without a callback.
2009-08-08 04:56:28 (r2623) by rcaputo
Move Philip Gwyn's signals test to POE::Test::Loops, so it can
exercise everyone's event loops.
2009-07-27 17:07:20 (r2608) by rcaputo
Reorganize documentation links.
2009-07-27 05:10:57 (r2605) by rcaputo
Added a machine-readable repository directory to the distribution.
Documented the bug tracker, repository, and other resource URLs.
momentous, it does not appear to be significant.)
Many minor fixes and enhancements. Notable entries:
- Added new Geosync tag to allow geotagging of images with timestamps which
are not pre-synchronized to GPS time
- Changed conversion of Canon MeasuredEV to correspond more closely to
LightValue (by adding 5 to the MeasuredEV value, which seems to be good for
all EOS models, but it may be high by up to 1 EV for some PowerShot models)
way to build sqlite3 according to upstream.
It should give a small performance increase due to static inlining,
but more importantly lets us avoid manual maintenance of sqlite3.h.
Bump revisions for sqlite3 and sqlite3-tcl.
Pkgsrc changes:
- Adjust dependencies
Upstream changes:
6.16 Sun Oct 11 08:57:18 GMT 2009
- BotTraffic.pm: Emit 'irc_bot_notice' events for bot notices (Hinrik)
- Logger.pm: Log NOTICEs if requested (Hinrik)
- Proxy.pm: Fix a regression introduced in 6.05_01. This was causing it
to be completely broken. Added a test so it won't happen again
unnoticed (Hinrik)
6.14 Thu Sep 24 15:07:05 GMT 2009
- More "return" -> "return PCI_EAT_NONE" fixes. This eliminates some
harmless (but annoying) warnings. (Hinrik)
- State.pm: Fix AwayPoll, which wasn't working at all. Also added a
test for it and made the documentation clearer. Thanks to David E.
Wheeler for spotting that one. (Hinrik)
- IRC.pm: Document which spawn() options can not be passed to the
'connect' event. (Hinrik)
- IRC.pm: Split PRIVMSGs with newlines into multiple messages. For other
commands, don't pass user-supplied newlines through to the IRC server
as it allows the user to submit raw IRC commands. (Hinrik)
6.12 Thu Sep 10 09:25:02 BST 2009
- Fix localaddr() issue reported in RT #48791 by Michael Andreen
- Depend on latest (1.24) POE::Component::Pluggable (Hinrik)
- BotCommand.pm: Strip colors/formatting before processing (Hinrik)
- Plugin::AutoJoin S_join should return PCI_EAT_NONE if
$joiner ne $irc->nick_name() (perigrin)
6.10 Fri Aug 14 21:19:07 BST 2009
- Implemented netsplit detection and handling of state on netjoin (bingos)
- Refactored the netsplit code for robustness and sanity (bingos)
- Added testcase for netsplit handling (bingos)
- AutoJoin.pm: Fixed problem with rejoining password-protected channels
that were not passed to the plugin constructor (Hinrik)
- Removed extended debug output from some tests, they've been behaving
for a while (Hinrik)
- State and subclasses will use NAMES replies to synchronise
channel state as well now. Should help RT #46825 (bingos).
- Refactored the netsplit test slightly to try and eliminate
race conditions (bingos)
- Added netsplit detection code to Qnet::State subclass. (bingos)
- Netsplit restoration now triggers irc_nick_sync event (bingos)
- Added some diagnostics to the netsplit test. (bingos)
- Markup test in netsplit as todo due to race condition (bingos)
- Mark one of the netsplit tests TODO (bingos)
- Time for a stable release
6.08 Fri May 29 11:46:45 GMT 2009
- CTCP.pm: Return an RFC822 date in response to CTCP TIME (Hinrik)
- BotCommand.pm: Fix RT #46065, help message wasn't being printed (Hinrik)
- Connector.pm: Make the traffic-noticing code more accurate (Hinrik)
- PlugMan.pm: Allow custom auth checks for the IRC interface (Hinrik)
- PlugMan.pm: Silence some warnings (bingos)
6.06 Thu Apr 30 12:05:04 GMT 2009
- NickServID.pm: Update a paragraph in the Pod (Hinrik)
- State.pm: Don't delete all state in S_(error|socketerr|disconnected)
handlers, removes some warnings (Hinrik)
- AutoJoin.pm: Fixed some bugs, added more tests (Hinrik)
- PlugMan.pm: Don't rely on State.pm for authentication. Eliminates race
condition when receiving channel commands before the channel has been
synced (Hinrik)
- In jailed environments we can't assume that 127.0.0.1 will be that.
reported by Jase Thew (Bazerka).
- Logger.pm: Replace slashes with underscores before logging to disk,
spotted by Sebastian Mair.
6.05_01 Sat Apr 11 09:18:28 GMT 2009
- Compat.pm: Don't emit an extra event or print a misleading debug message
for CTCP ACTIONs on FreeNode (Hinrik)
- AutoJoin.pm: Only join channels after we have asked the server if it
supports FreeNode's CAPAB IDENTIFY-MSG (Hinrik)
- BotCommand.pm: Accept commands in private too (Hinrik)
- CTCP.pm: Handle CLIENTINFO as well (Hinrik)
- Common.pm: Added irc_to_utf8 to decode IRC messages (Hinrik)
- Proxy.pm: General cleanup. Also fix a bug introduced in 5.66 that caused
it to keep too many welcome messages (Hinrik)
- State.pm: General cleanup. Fixed a bug with the order of irc_nick_sync's
arguments being reversed (Hinrik)
6.04 Sat Mar 7 23:31:11 GMT 2009
- Logger.pm: Only use portable strftime parameters. Fixes log timestamps
on Solaris and Windows (Hinrik)
- CTCP.pm: Use portable strftime parameters when responding to CTCP TIME
requests (Hinrik)
- State.pm: Document the extra parameters to irc_disconnected, irc_error,
and irc_socketerr. (Hinrik)
- Fixed race condition in some tests (Hinrik)
6.02 Fri Mar 6 10:54:22 GMT 2009
- Fix RT #43856, variable name typo in PlugMan.pm reported by barnaclebob
- Add new test for PlugMan plugin (Hinrik)
- Amend a few tests that were failing (Hinrik)
- State.pm: Avoid a warning when setting a mode on a channel on which all
previous modes have been unset (Hinrik)
- Add optional Perl::Critic test for the test scripts (Hinrik)
6.00 Wed Mar 4 23:12:57 GMT 2009
- Logger.pm: Fix bug with logging some CTCP ACTIONs (Hinrik)
- Logger.pm: Also log own messages in DCC chats. Add test for it (Hinrik)
- Added more tests for general IRC activity, and for State.pm (Hinrik)
- Added tests for BotCommand and Logger plugin (Hinrik)
- Turned off flood control in all the tests, and removed or adjusted many
of the delays used. This shaves about 70% off the time needed for a full
test suite run (Hinrik)
5.98 Mon Mar 2 22:51:27 GMT 2009
- Parse some IRC protocol messages more strictly. Always split on ASCII
space rather than \s, since tabs are not considered whitespace by the
IRC protocol. (Hinrik)
- DCC.pm: Most events now tell you what the peer's IP address is (Hinrik)
- DCC.pm: Document the timeout parameter to the dcc command, and fixed
some errors in the docs. (Hinrik)
- Compat.pm/DCC.pm: Actually provide the whole nick!user@host (not just
the nick) with every dcc_request, like the synopsis suggests (Hinrik)
- DCC.pm: Fix crash when closing a DCC connection with pending outgoing
data, reported by meneldor (Hinrik)
- Added test case for the above (Hinrik)
- DCC.pm: Don't crash if dcc_close is called with an invalid id, also
reported by meneldor (Hinrik)
- Logger.pm: Add DCC chat logging (Hinrik)
- IRC.pm: Document the 'account' key returned by irc_whois (Hinrik)
- IRC.pm: Allow plugins to respond to custom commands without them having
to be defined explicitly in IRC.pm. This allows the removal of the last
piece of DCC-specific code (Hinrik)
- AutoJoin.pm: Add an option for retrying joins when banned (Hinrik)
- NickServID.pm: Support ratbox-based ircds (Hinrik)
- Synchronised all the version numbering (bingos)
5.96 Wed Jan 28 11:29:28 GMT 2009
- Added delays to the two failing tests reported by CPAN Testers,
believe we are seeing race conditions. (bingos)
5.94 Tue Jan 27 21:38:51 GMT 2009
- Fixed dependency on Date::Format in inc. poco-server-irc (bingos)
5.92 Tue Jan 27 13:18:12 GMT 2009
- PlugMan.pm: Only require ::State when 'botowner' is set (Hinrik)
- ISupport.pm: Fix parsing of MODES and SILENCE parameters (Hinrik)
- AutoJoin.pm: Added 'Rejoin_delay' option (Hinrik)
- Connector.pm: Allow adjusting the time to wait before reconnecting,
to ease testing. (Hinrik)
- Compat.pm: Fix parsing of CTCPs when no prefix is present
(i.e. client CTCPs) (Hinrik)
- Updated included POE::Component::Server::IRC to 1.36 (Hinrik)
- Fixed up some tests to work with it (Hinrik)
- Removed ziplink test since PoCo-Server-IRC only supports server
ziplinks, not client ones (Hinrik)
- Added tests for the following plugins: AutoJoin, BotAddressed,
CycleEmpty, CTCP, Connector, ISupport, NickReclaim, Whois (Hinrik)
5.90 Thu Jan 22 10:52:53 GMT 2009
- Seen.pod: Recipe for a bot implementing the 'seen' command (Hinrik)
- Reload.pod: How to reload your bot with out reconnecting (Hinrik)
- Memory leak with stashing $self in $self->{alias}, change to
stash a stringified version of $self instead. (bingos)
5.88 Thu Aug 28 15:49:48 BST 2008
- MegaHAL.pod: Bare-bones recipe for a MegaHAL bot (Hinrik)
- BotCommand.pm: Send responses back via NOTICE, not PRIVMSG (Hinrik)
- Filter/CTCP.pm: Removed, as it is deprecated and unmaintained (Hinrik)
- IRC.pm: Really propagate the plugin_debug flag (Hinrik)
- Filter/IRC/Compat.pm - change to _get_ctcp() as per RT #38773
pkgsrc changes:
- Adjusting license definition
Upstream changes:
1.22 -- Sat Nov 14 09:26:15 CST 2009
* silence a compiler warning about an unreferenced local variable [Steve Hay]
* RT#51484 Preserve utf8 flag of string passed to dualvar()
* RT#51454 Check first argument to first/reduce is a code reference
* RT#50528 [PATCH] p_tainted.t fix for VMS [Craig A. Berry]
* RT#48550 fix pure perl looks_like_number not to match non-ascii digits
pkgsrc changes:
- Adjusting dependencies according to META.yml and upstream changelog
- Ensure that Module::Autoinstall won't fire
Upstream changes:
3.09000 2009-11-17
- Bumped Net::OAuth requirement to 0.20 (hopefully resolves CPAN Testers failures)
- Added users_search/find_people API method
- Updated default apiurl: http://api.twitter.com/1
- Deprecated apihost option to new
- Fixed ssl handling for Lists API
3.08000 2009-11-02
- return user_id and screen_name from request_access_token (Roberto Etcheverry)
3.07999_01 2009-10-23
- Added experimental Lists API support
3.07004 2009-10-22
- Fixed: cached request_token results in a stale authentication/authorization urls
3.07003 2009-10-13
- Added new API method: report_spam
3.07002 2009-10-05
- Documented the cursor parameter for friends, followers, friends_ids, and
followers_ids.
- Fixed: synthetic since parameter (could fail with multiple API traits defined)
- Added support for since in YYYY-MM-DD format
3.07001 2009-09-28
- Added specific versions for DateTime and DateTime::Format::Strptime (closes RT#50069)
- better deps (Alexandr Ciornii)
3.07000 2009-09-22
- added 'retweets' REST API method
- added synthetic 'since' parameter
- fixed: object inflation for Search API returns (different timestamp format!)
pkgsrc changes:
- Adjusting dependencies according to META.yml (see Upstream changes)
- Setting Module::Build as build type
Upstream changes:
0.20 Fri, 13 Nov 2009 18:56:55 UTC
- Fixed http://rt.cpan.org/Public/Bug/Display.html?id=48867 - error in
synopsis - thanks Adam Taylor!
- Removed UTF8 double-encoding warning. Now Net::OAuth::Message::encode()
uses Encode::is_utf8() to determine if the input is already UTF-8 encoded.
If so, it runs decode_utf8() on it before sending it to uri_escape_utf8().
Thanks Hector Garcia Alvarez!
- Potentially fixed an issue found by Marc Mims, where HMAC_SHA1 was failing
to load, breaking some CPAN tests. Build.PL now explicitly requires
Digest::SHA1 and Encode. Hopefully that fixes it.
pkgsrc changes:
- Adding license definition
- MIME::Base is included in perl core in high enough version
Upstream changes:
0.08 Do 29. Okt 22:09:12 CET 2009
- added LWP::Debug to dependencies...
0.07 Di 27. Okt 20:51:50 CET 2009
- fixed <https://rt.cpan.org/Public/Bug/Display.html?id=50881>
Thanks to Adam Sjogren.
pkgsrc changes:
- adjusting license definition
- adjusting dependency to Test::More (0.72 is in core, 0.62 is required)
Upstream changes:
0.11 Tue Nov 10 14:14:17 EST 2009
No code changes; add Win32::Event dep for windows platforms
Upstream changes:
0.13 Fri Oct 9 15:01:07 EDT 2009
* Better cleanup temp files after tests
* No longer add a double / in the path when testing for index.html
* Better support for sending custom HTTP statuses in $m->abort
pkgsrc changes:
- Added license definition
- Removed test patch (test runs fine without)
Upstream changes:
2.2 Thu 1 Oct 2009
- Moved Lite.pm into lib directory
- Reversed the Changes file order to reverse chronological
- Moved to Makefile.PL to Module::Install to autodetect more things
- Add no_index entries for the test data
- Remove reliable on (partially) dead website for testing
- Now this module is mature and unlikely to change much, move to
a more stable and less complicated versioning scheme that suffers
less complications in the Perl toolchain. Also, dotted integers
without the use of something like version.pm isn't strictly
allowed.
