[Boost-users] Build error

30 views
Skip to first unread message

OvermindDL1

unread,
Jul 14, 2009, 1:31:30 AM7/14/09
to boost...@lists.boost.org
I am attempting to build boost, first I set some environment vars
(some may not be useful anymore, but I still set them):
set BZIP2_SOURCE=G:\SDKs\boost\bzip2-1.0.4
set ZLIB_SOURCE=G:\SDKs\boost\zlib-1.2.3
set HAVE_ICU=1
set ICU_PATH=R:\SDKs\icu\icu
set EXPAT_INCLUDE=G:\SDKs\expat\expat_cvs\expat\lib
set EXPAT_LIBPATH=G:\SDKs\expat\expat_cvs\expat\win32\bin\release

I then call:
bootstrap.bat
bjam --build-type=complete --toolset=msvc-8.0 --without-mpi
--without-python --prefix=R:/SDKs/boost/built_head
--build-dir=R:/SDKs/boost/build_head define=_CRT_NONSTDC_NO_DEPRECATE
define=_CRT_SECURE_NO_DEPRECATE define=_SECURE_SCL=0
define=_SCL_SECURE_NO_DEPRECATE define=_HAS_ITERATOR_DEBUGGING=0
install>..\build-HEAD.log

And my build-HEAD.log contains:
Note: Building Boost.Regex with Unicode/ICU support enabled
Using ICU in R:\SDKs\icu\icu/include
warning: Graph library does not contain MPI-based parallel components.
note: to enable them, add "using mpi ;" to your user_config.jam
error: link=shared together with runtime-link=static is not allowed
error: such property combination is either impossible
error: or too dangerious to be of any use

I have never seen those last three errors, and other then the
bootstrap my commands have not changed in many many versions. What is
causing them?
_______________________________________________
Boost-users mailing list
Boost...@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users

Vladimir Prus

unread,
Jul 14, 2009, 3:00:21 AM7/14/09
to boost...@lists.boost.org
OvermindDL1 wrote:

I think this matter was raised on this mailing list before -- IIRC,
ICU has only shared version, so trying to mix it with runtime-link=static
is not going to work. I suggest you don't use --build-type=complete, and
explicitly request the variants you want -- and don't request static runtime.

HTH,
Volodya

OvermindDL1

unread,
Jul 14, 2009, 4:16:21 AM7/14/09
to boost...@lists.boost.org

Hmm, I thought I looked over every email on the list since I
subscribed however long ago, guess I missed it or it did not stick in
my mind.

So ICU is at fault then, I shall test that by just removing ICU from
my build, I last needed it like a year ago for an old project, nothing
recent (well, nothing needing it through Boost anyway, although I do
still need it). I am now doing the exact same build as above, but
removing the HAVE_ICU environment var.

Yep, that was it, the build log is now filling up faster then I can
keep up, thanks.

I am curious though, the last build I did of Boost was the trunk only
about 3 or 4 months ago, and it was successful just fine then, so
something changed only recently. I have never really looked at the
bjam build scripts so I do not really know where to look either, but
should not the Boost.Regex bjam script disallow that combination from
even appearing in a complete build? Sounds like a simple fix for
someone who knows the bjam scripts.

While I am at it, I am curious as to what I could put in my
project-config.jam file that would fulfill everything I normally do
(setting the environment vars, and everything I add on the bjam
command line) so all I would need to do is bootstrap then bjam?

Vladimir Prus

unread,
Jul 14, 2009, 7:55:58 AM7/14/09
to boost...@lists.boost.org
OvermindDL1 wrote:

I suggest you file a bug at svn.boost.org

> While I am at it, I am curious as to what I could put in my
> project-config.jam file that would fulfill everything I normally do
> (setting the environment vars, and everything I add on the bjam
> command line) so all I would need to do is bootstrap then bjam?

You can put this in project-config.jam:

project project-config : default-build <variant>release <threading>single ;

to specify default properties for a build. The properties I gave above is
an example -- you can put anything there. It is not presently possible to
put --build-type there. I am not sure what environment variables you mean.

- Volodya

Reply all
Reply to author
Forward
0 new messages