Re: [Boost-testing] out of date results for release branch

Subject: Re: [Boost-testing] out of date results for release branch
From: Eric Niebler (eric_at_[hidden])
Date: 2010-06-25 23:45:26

On 6/25/2010 11:29 PM, Jim Bell wrote:
>> I appreciate your work. I wasn't criticizing any of the testers. ...
> 
> No offense taken. I appreciate your frustration not knowing where a change's
> tests stand.
> 
>>> ... why wouldn't a test result be valid until the next release, ...
> 
>> The release branch is constantly getting patches merged from trunk. ...
> 
> I didn't realize that. But a particular version (say, boost-1.43.0) is the
> same whether you download it the day it's released or ten years from now,
> right? 

Right.

> Is there a mechanism for fetching, say, boost-1.43.1?
> 
> Just curious.

Boost doesn't do point releases. That's not what the release branch is
for. Boost has two branches: trunk and release. Trunk is the Wild West:
developers make changes there, stuff breaks often, they fix it. When it
looks good enough to be in the next release, they merge their changes
from trunk to release. This happens continuously until, at 3 month
intervals, we freeze the release branch, stabilize it, tag it and
release it.

Breakages happen rarely on the release branch because changes have been
tested first on trunk, but it still happens and we need to know about
it. That's why we need testers on both trunk and release to run tests
continuously.

HTH,

-- 
Eric Niebler
BoostPro Computing
http://www.boostpro.com