One small step for mankind - was Re: What's happened to Ryppl?

Dave Abrahams wrote:
Robert, great questions. Could you post them to the boost developers' list? They really don't belong here.
Thanks.
On Sat, Jan 29, 2011 at 12:57 PM, Robert Ramey <ramey@rrsd.com> wrote:
Dave Abrahams wrote:
On Fri, Jan 28, 2011 at 4:35 PM, Robert Ramey <ramey@rrsd.com> wrote:
In the interest of moving toward a better modularization of boost and decoupling of libraries, I would like to make a suggestion:
Can we change library testing so that each library is tested against the current release branch of all the other libraries?
When I test on my own machine, I don't test against the trunk. I test against the current release tree. Another way of saying this is that on my current machine I have my directory tree set to the the boost release branch. ONLY the serialization library directories are set to the trunk. So when I test on my local machine I KNOW that when I merge into the release branch I won't have any unexpected problems.
a) I don't think this would be a huge change. b) It would better test what the user does c) It would isolate each library from another so that errors in the trunk (wild west) in one library don't impact on development of other libaries. d) It would promote decoupling of libraries. e) as an intermediate step, not all testers would have to make the change - some could use the old script.
It would be a small, not too difficult step on the path we think we want to travel.
Robert Ramey

At Sat, 29 Jan 2011 15:35:10 -0800, Robert Ramey wrote:
Dave Abrahams wrote:
Robert, great questions. Could you post them to the boost developers' list? They really don't belong here.
Thanks.
On Sat, Jan 29, 2011 at 12:57 PM, Robert Ramey <ramey@rrsd.com> wrote:
Dave Abrahams wrote:
On Fri, Jan 28, 2011 at 4:35 PM, Robert Ramey <ramey@rrsd.com> wrote:
In the interest of moving toward a better modularization of boost and decoupling of libraries, I would like to make a suggestion:
Can we change library testing so that each library is tested against the current release branch of all the other libraries?
When I test on my own machine, I don't test against the trunk.
Oh... the release *branch* rather than the *released version*. Interesting. +1. However, we should be prepared: that could make for a lot of churn on test machines. I think Boost's testing focus should be on 1) release candidates (i.e. what's in the release branch today) and secondarily on 2) libraries in-development as tested against the release candidates of everything else. I don't think there's any reason to show all developers the complete results of #2 unless they want to go out of their way to see them; the only #2 results I care about are those testing my libraries. -- Dave Abrahams BoostPro Computing http://www.boostpro.com
participants (2)
-
Dave Abrahams
-
Robert Ramey