Almost missed this email. Sorry about the delayed response..

On Sun, Oct 16, 2016 at 6:27 PM, Stefan Seefeld <> wrote:
On 16.10.2016 18:36, Rene Rivera wrote:
> On Sun, Oct 16, 2016 at 5:07 PM, Steven Watanabe <
> <>> wrote:
>     On 10/16/2016 02:55 PM, Stefan Seefeld wrote:
>       Is there any reason you can't just
>     call sphynx-build directly from the Jamfile
>     without going through make?
> Because that doc building should only run for the "boostrelease" target.

Can you elaborate ? I don't understand how the target that invokes the
doc build relates to how the doc build is implemented.

It only relates in that.. Only two targets are considered (i.e. invoked) when building the documentation for releases (boostdoc and boostrelease).
For simplicity's sake, right now I'd like a simple invocation of
`.../b2` in the 'doc/' directory to build all of Boost.Python's

That's fine for "standalone" docs. But not for the release. Just running a build in the doc directory is not enough to discriminate what documentation needs to get built for a release. As some libraries support all three modes of building: standalone (ie as if outside the Boost tree), fully integrated (shows up in the global doc book in root/docs/html), and non-integrated (ie building standalone but linking as if in the release archive.
A related question I was wondering about: is the documentation building
being CI-tested somehow & somewhere ?

Yes. Here:


Although I'm going to remove the release building from Travis as their limits are currently making the build fail.
Can the generated docs be
inspected for manual / visual testing ?

Two options..

1. Download the snapshot builds as they are uploaded directly from the CI servers at SF <>.

2. Those same builds are used to post docs tot he website at the following depending on the branch:


-- Rene Rivera
-- Grafik - Don't Assume Anything
-- Robot Dreams -
-- rrivera/ (msn) - grafikrobot/aim,yahoo,skype,efnet,gmail