2009-12-19 17:01:44 -0500 v1_163

commit 27b7665de4fa1e3b3960720ddd83632f21c60497 Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Dec 19 17:01:44 2009 -0500

Avoid "UNKNOWN" reports from CPAN testers. All tests were skipped when not tested as root, so add a test for non-root users.


2009-10-14 18:00:18 -0400 v1_162

commit 65682a45ea5d0ecd93625389e4fe2372334bd3f7 Author: Rocco Caputo <rcaputo@cpan.org> Date: Wed Oct 14 18:00:18 2009 -0400

A test was failing rather than skipping due to insufficient privileges.


2009-09-12 18:56:44 -0400 v1_161

commit 850b00fc1e9f70833ab36ac13a2b38d134e2c6af Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Sep 12 18:56:44 2009 -0400

Standardize layout and prepare for next release.

commit 420df6d9d9345a5ed294eb4fec88bf14d8a41266 Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Sep 12 18:48:25 2009 -0400

Add test for timeouts inflating retried response times.

commit 6cd8830b9f822296ecbe7db0633502d9cf4e3195 Author: Ralph Schmitt <ralph.schmitt@ish.de> Date: Sat Sep 12 18:14:32 2009 -0400

Compensate for timeout when retrying.

Ralph Schmitt discovered that the response time each retry would increase by the timeout period. He provided this change to compensate.

commit 1ed1c1da115bbd7b8e63eff439225c5fe6d664ff Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Sep 12 17:12:28 2009 -0400

Convert files and release workflow from svn to git.


2009-07-28 06:31:09 +0000 v1_160

commit 96f372e227aa190a221394805eb9d6ae76fdb7ee Author: Rocco Caputo <rcaputo@cpan.org> Date: Tue Jul 28 06:31:09 2009 +0000

Use latest POE, and prepare for a new release.

commit 24380582236a0354365aa68d0709e5340c52abff Author: Rocco Caputo <rcaputo@cpan.org> Date: Mon Jul 27 04:56:40 2009 +0000

Added a machine-readable repository directory to the distribution. Documented the bug tracker, repository, and other resource URLs.


2009-07-25 17:26:55 +0000 v1_15

commit 1228b4e483109d37bd49193479fdc85f866401bd Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Jul 25 17:26:55 2009 +0000

Resolve rt.cpan.org ticket 35738. Isaacson rightly pointed out that the component would never end if an application gave it its own raw socket. Now the component will stop watching the socket when there's nothing to do, which will allow it to exit if the rest of the program is also stopping. Added a test case for the condition, too.

commit 8ea6ea61ca8fda54612f1964ff028e7d982f290d Author: Rocco Caputo <rcaputo@cpan.org> Date: Sat Jul 25 07:45:33 2009 +0000

Calculate a ballpark optimal Parallelism value by default. Previously, the component would try to process as many simultaneous requests as possible, often exceeding OS and network limits, and either dropping packets or returning responses with patently bogus round-trip times. Thanks to Dean Holroyd for rt.cpan.org #48171, which this should resolve.


End of Excerpt