Boost's authors have created a set of top-notch libraries. In return, they're inundated with bug-reports ("tickets") of all sorts. And as the libraries' regression tests are run on more and more platforms, even maintaining the tests themselves is more of a burden than the authors can maintain.

Finding some unexpected issues with the boost libraries (I'll tell my story eventually), I found myself doing my own audit of boost's release and bug-fix mechanisms.

There's way too much yellow (unexplained failures) on the regression tests (both release and trunk). And there are too many open tickets.

Boost needs two sets of new volunteers to attack the Tickets and regression tests. The authors and release managers have more to do than they can handle. A fresh set of hands (and minds) needs to attack it.

Here are my ideas to address it.

Special projects that would help the Guild (and that the Guild could help with):

  • Diffs between successive regression tests for (libX, testY, platformZ).
  • Sandbox maintenance. Search for Sandbox on the Nov-2010 boost list. (Though I hope the Guild can recruit enough effort to handle this, its first priority is Boost proper.)

Feedback welcome.

Status update, 2/12/2011: Boost.Guild's showstopper is in finding maintainers to commit changes back to the trunk (and carry them forward to release). I'm watching the boost.pool hit-squad effort to see how that's going. (Group here.)


Status update, 12/18/2013: A big step forward in this realm, with the creation of Boost's Community Maintenance Team.