Re: [jamboost] Re: Build properties algorithm

From: Zbynek Winkler (zwin_at_[hidden])
Date: 2004-04-02 10:44:32

Here is some more info.

In the definition of 'lib cppunit' there was missing <threading>multi. 
After adding it, it went fine. So instead of using

lib cppunit : : <name>cppunitd60 <variant>debug 
<toolset-msvc:version>6.0 <search>d:/software/cppunit-1.8.0/lib ;

I used

lib cppunit : : <name>cppunitd60 <variant>debug <threading>multi 
<toolset-msvc:version>6.0 <search>d:/software/cppunit-1.8.0/lib ;

The algorithm was said to handle <toolset>msvc-6.0 instead of 
<toolset-msvc:version>6.0 so I tried

lib cppunit : : <name>cppunitd60 <variant>debug <threading>multi 
<toolset>msvc-6.0 <search>d:/software/cppunit-1.8.0/lib ;

but that did not work. For some (to me unknown) reason bjam tried to 
link against cppunit71.lib so I've got bunch of unresolved referencies.

I use msvc-6.0 as my default toolset so I tried compiling with msvc-7.1 
by issueing 'bjam toolset=msvc-7.1' and this is what I've got:

d:/software/boost-build-cvs/v2/build\feature.jam:680: in 
expand-composites from module feature
error: explicitly-specified values of non-free feature 
<toolset-msvc:version> conflict
existing values: 7.1 6.0

For some (to me unknown) reason using 'bjam toolset=msvc-6.0' works... 
Using 'bjam toolset=msvc toolset-msvc:version=7.1' does not help either. 
The most curious thing is that using

lib cppunit : : <name>cppunitd60 <variant>debug <threading>multi 
<toolset>msvc-6.0 <search>d:/software/cppunit-1.8.0/lib ;
lib cppunit : : <name>cppunitd71 <variant>debug <threading>multi 
<toolset-msvc:version>7.1 <search>d:/software/cppunit-1.8.0/lib ;

in my user-config.jam I am able to compile and link programs against the 
cppunit library. However when I change the second definition of cppunit to

lib cppunit : : <name>cppunitd71 <variant>debug <threading>multi 
<toolset>msvc-7.1 <search>d:/software/cppunit-1.8.0/lib ;

I get the error I've started with about no best alternative but when 
compiling with *msvc-6.0*. Compilation with mingw works without problems 
(just as before the upgrade) so the problem could be related to the 
subfeature handling.

I give up.

Zbynek

PS. I suggest adding some unit test for this so we do not get such 
reggressions in the future.

-- 
http://zw.matfyz.cz/ http://robotika.cz/
Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic