Re: [Boost-build] Boost 1.35.0 Jamroot patch to make install work n subprojects

From: Steven Watanabe (watanabesj_at_[hidden])
Date: 2008-08-31 00:10:30

AMDG

Jurko Gospodnetić wrote:
>   Anyone played around with this patch yet? Anyone have something 
> committable?

I've tested the attached patch locally.

* prefix header glob paths with $(BOOST_ROOT) so that they will be
  correct when bjam is run from a different directory.
* ditto for <install-source-root>
* replace nonexistent /boost/install-headers with 
/boost/install-proper-headers
  in boost-install.  (This target was renamed in [35986])

>   Here are some of my comments on it:
>
>   * Rule boost-install() seems to have an echo call leftover from some 
> debugging/development effort.

Oops.  I've removed it.

>   * What is the install-proper-headers target? Or more precisely, what 
> is the install-proper stage target and how does it relate to the 
> install stage target created in the boost-install() rule.

The install-proper does all the actual installation.
If the install target created by boost-install actually worked,
it would install all the headers and a single library.

>   * Not sure if your question 'where does the install-headers target 
> get created' is still an active one but as I see it it gets created in 
> the very same package.install() call that uses it.

I was referring to the use of /boost/install-headers which does not in 
fact exist.
The package.install creates something more like 
/boost/filesystem//install-headers

>   If this is a valid patch that actually does solve a real usability 
> problem with the Boost library's build, I'd hate to see it fall 
> through the cracks.

Another related problem is to make
bjam msvc -n --with-iostreams trunk//install
work, where trunk is my working copy.
After applying this patch, only the default-build
is still wrong.

In Christ,
Steven Watanabe