Re: [Boost-build] Changed behaviour of custom generators?

Subject: Re: [Boost-build] Changed behaviour of custom generators?
From: Steven Watanabe (watanabesj_at_[hidden])
Date: 2014-04-03 11:02:52

AMDG

On 04/03/2014 05:19 AM, Roland Schwarz wrote:
> On 02.04.2014 17:10, Steven Watanabe wrote:
>>
>> Can you try running with -d+4 which should print
>> out the string passed to CreateProcessA.
> Ok, several things happened:
> 1)
> The log entry changed to "Command string for CreateProcessA():
> '-ftemplate-depth-128  ..."
> which I know from earlier experience as indicating an inability of
> boost-build to detect my compiler.

I suppose this is a b2 bug.  It shouldn't
stop dead when an updating action fails.

> In my user-config.jam it is defined as:
> 
> using gcc
>     : 4.6.2
>     : "C:/subsys/win32/mingw-4.6.2/bin/g++.exe"
>     : <flavor>mingw32
>       <root>"C:/subsys/win32/mingw-4.6.2/"
>       <archiver>"C:/subsys/win32/mingw-4.6.2/bin/ar.exe"
>       <rc>"C:/subsys/win32/mingw-4.6.2/bin/windres.exe"
>       <cxxflags>-fno-strict-aliasing
>       <linkflags>-Wl,--allow-multiple-definition
>     ;
> 
> And I request this compiler from the cmd line as:
> b2 mytarget gcc-4.6.2-mingw32
> which worked before, and still works for the 4.7.2 compiler which is
> also in user-config.jam (altough it is the first compiler specified in
> the file, the default compiler)
> 

Can you double check that the command matches
the toolset definition.  I've tried quite a
few variations, and the only way I could
replicate the error was to use something like:

using gcc : 4.6.2 : ... : <flavor>mingw32 ;
using gcc : 4.8.1 : ... : <flavor>mingw64 ;

> b2 gcc-4.6.2-mingw64

Can you run b2 --debug-configuration --debug-building ...
with the failing toolset and give me the output?
It'll produce a lot, so you can restrict it to a
single target.

> 2)
> A second observation is, that I am not able any-more to use
> toolset-add-requirements within the user-config.jam to add requirements
> (well indeed only a specific one) to my compilers

This is a known bug.  I fixed it in
http://github.com/boostorg/build/commit/b869e58fa036ee2eec396dded5266d0a18c31ff5

In Christ,
Steven Watanabe