Subject: Re: [boost] Master Regression Tests Failing
From: Robert Ramey (ramey_at_[hidden])
Date: 2017-10-10 04:59:36
On 10/9/17 9:47 PM, Stefan Seefeld via Boost wrote:
> Indeed. Though, I agree with Robert and would go one step further: Given
> that the things under development are the boost libraries, not the tools
> used to build and test them, I would prefer to use a stable Boost.Build
> release, even when building and testing Boost's "develop" branch.
LOL - I go even further. On my local system I
a) clone the boost master branch
b) build boost tools, etc ....
c) pull the boost serialization library or any library I'm working on
d) switch the library under test (ie the serialization library in this
case to the develop branch
e) move the serialization/test directory
f) run b2 with all the switches
Now I'm testing my local develop version against everything else on the
master. Now I know that
a) any errors are my own and not some artifact from someone elses work
on the develop branch.
b) When I merge my develop branch to the master, I am almost sure to not
have any errors when the master branch tests are run.
Makes my life much, much simpler.
And best of all, I don't have to convince anyone else to go along. I
can just do it in the privacy of my own home and no one is the wiser.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk