Build, Release and QA --------------------- *Tinderbox debugging* Dan and John proposed loading dev tools on tinderboxen: wing on all platforms, Microsoft Visual Studio on windows, GDB on other platforms. They also proposed running full debug tests with the --wing option. The tests would run more slowly, but would help us track down hard to reproduce crashes when we encounter them on tinderboxes. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007293.html
Bryan Stearns suggested that running with --wing would be a bad idea, hanging the tinderbox for any exception caught. Bryan asked if it would be helpful to install tbox software on a developer's machine. John suggested configuring wing so that it didn't stop on exceptions, or at least installing wing. John noted that he's been running functional tests on his computers at night, allowing him to find and fix rare failures. He asked others to try it out. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007295.html *Alpha4 RC* We ran into some new bugs, consequences of the defer-delete change. Philippe asked dogfooders to continue using the most recent checkpoint. http://downloads.osafoundation.org/chandler/checkpoints/ He asked developers to act promptly on alpha4 bugs. (We ended up spinning the first release candidate on Wednesday.) http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007296.html *Checkpoints and RC* Bear spun a branch checkpoint on Monday and an RC on Wednesday. *QA Summary* Aparna gave a summary of the QA team's activities: http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007300.html *Eggs and moving icons* Bear moved chandler/icons into chandler/Chandler.egg-info/resources/. *0.7 docs on Chandler site* Pieter and Katie agreed that we should start pointing at the 0.7 developer docs on the Chandler site (docs/0.7/ instead of docs/0.6/). http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007311.html *SMTP woes* Jeffrey and others discovered problems with tinderbox tests using SMTP. Reid suggested the problem might be non fully qualified domain names, as this is how he recently fixed his OSAF bounces. Reid asked if it was appropriate for the server to be strict here or for the clients to adapt. PJE pointed out that most servers are configured this way, and that valid FQDN host names have been a required part of the protocol for many years. Jared confirmed that OSAF recently changed the servers (increasing standards-compliance) as part of a de-spamification effort. Jared also pointed out a recent domain name typo in one of the tests, which Dan fixed. Thread starts here: http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007313.html Dev Projects ------------ *defaultValue for displayName* Bryan ran into snags adding a defaultValue to ContentItem's displayName: 'subject' is special cased in sharing code, changing it would be a sharing format change, and a test expects mail messages to have a 'subject' attribute. Bryan doesn't need this change for alpha5, so he's going to move on for now. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007297.html Morgen asked if 'subject' used to be a real attribute, and wondered if we're only maintaining it in the sharing XML for backwards compatibility. Grant explained that the direction of the redirection changed. (MailMessageMixin used to redirect the generic 'about' to its own 'subject'. Currently MailStamp's 'subject' redirects to ContentItem's 'displayName'.) Grant suggested taking advantage of the backward compatibility break in alpha4 -- never write out redirected attributes. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007304.html Morgen summarized observations and decisions made in the IRC: - Assigning a defaultValue to displayName breaks sharing code that maintains backward compatibility. Removing that code would cause instability for alpha4, and break existing shares. - The defaultValue isn't required, we can change all code that relies on getItemDisplayName() to provide a default in the code. - We can reasses when we move to EIM-based sharing. - Morgen, John and Bryan are removing all calls to getItemDisplayName(). - Andi can remove getItemDisplayName() once they are done. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007307.html Meetings and Announcements -------------------------- *Improving functional tests* Philippe summarized a meeting about improving functional tests. John will be spending 2 weeks working with the QA team to improve QATestLib. http://lists.osafoundation.org/pipermail/chandler-dev/2006-November/007309.html _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Open Source Applications Foundation "chandler-dev" mailing list http://lists.osafoundation.org/mailman/listinfo/chandler-dev
