PEAR 1.4.1 release testing

Hi all, I've included pecl-dev in this message as most of the issues fixed affect PECL users more directly than PEAR users. If any of you can drag even a moment away to try out the new release, that would be very helpful for ensuring the fixes I've put in actually work. PECL devs can feel free to ignore the .phars, I am more interested in the .tgz Due to some pressing issues in PEAR 1.4.0, I've gone ahead and prepared for the next release cycle. Testing tarballs are available at: http://pear.php.net/~greg requested testing: $ pear upgrade http://pear.php.net/~greg/PEAR-1.4.1.tgz - install a PEAR package - upgrade a PEAR package - uninstall a PEAR package - uninstall a package that has another depending on it to make sure this fails - remote-list/list-all/remote-info/list-upgrades - info/list if you can, checkout pear-core/tests/ and $ pear run-tests -rq This release is going out on Sunday unless serious issues are found. There is 1 test that fails on my system, but I don't see anything actually wrong with the output (it's a weird failure), so I don't think it is a real issue. [1] This release is important because it fixes severe issues such as: 1) pecl packages with package2.xml configure_options will get a fatal error on installation 2) packaging with 2 package.xml does not test for equivalency strictly enough 3) invalid php dependencies in package.xml 1.0 were not being flagged as invalid 4) "pear convert" doesn't work for extension source packages with install-as/platform attributes [this applies directly to the pecl_http package] There's others, it's all in the release notes. If any of you can test go-pear.phar and verify that it works on linux/windows in PHP 5.1+, and test the install-pear.phar with a checkout of the PHP_5_1 branch (put it in php-src/pear/install-pear.phar and run "make install su"), that would be good too. Greg [1] The test has 2 errors whose error messages differ so slightly, I can't even see the difference. I suspect it is not matching some whitespace or something, but the expected errors are still being thrown and it looks the same, so in my book it passes, and the testing framework is what fails here :).