|
Boost-Build : |
From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2019-12-17 11:14:05
On Mon, 16 Dec 2019 at 22:35, Mateusz Loskot <mateusz_at_[hidden]> wrote:
>
> On Windows, I'm building tests of Boost.GIL using
> dependencies like libpng, libjpeg, etc. installed with vcpkg.
>
> In my user-config.jam I have entries like this:
>
> using libjpeg
> :
> : <include>C:/vcpkg/installed/x86-windows/include
> <search>C:/vcpkg/installed/x86-windows/lib
> : <address-model>32
> ;
>
> using libjpeg
> :
> : <include>C:/vcpkg/installed/x64-windows/include
> <search>C:/vcpkg/installed/x64-windows/lib
> : <address-model>64
> ;
>
>
> At build time, everything works fine.
>
> For run-time, GIL tests need to load DLLs in C:\vcpkg\installed\x86-windows\bin
> If I add this location to PATH manually, everything works at run-time fine.
>
> I'm looking for a way to add that location to PATH from user-config.jam.
> Is it possible? How?
I approached the problem differently and added this at the top of my
user-config.jam
project
: requirements
<address-model>32:<dll-path>C:/vcpkg/installed/x86-windows/bin
<address-model>64:<dll-path>C:/vcpkg/installed/x64-windows/bin
;
Now all the DLLs of libjpeg, libpng, etc. can be found when I run
either of these two:
b2.exe toolset=msvc-14.2 address-model=32 libs/gil/test/extension/io//simple
b2.exe toolset=msvc-14.2 address-model=64 libs/gil/test/extension/io//simple
Spectacular success of mine! ;)
Next, I'd like to make this work for request of both at the same time,
that is address-model=32,64 or address-model=64,32
Interestingly, both first target in the list succeeds and the second
fails with linking
errors due to mismatch of machine type of libraries and build target.
That is, unresolved external symbol errors are described as follows
- for address-model=32,64
warning LNK4272: library machine type 'x86' conflicts with target
machine type 'x64'
- for address-model=64,32
warning LNK4272: library machine type 'x64' conflicts with target
machine type 'x86'
For both address models, checks of libraries succeed with yes.
Is there any way to make the multi-model command work?
Best regards,
-- Mateusz Loskot, http://mateusz.loskot.net
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk