On Friday, August 13, 2010 11:10 AM, Johan Råde wrote:
Thank you for a detailed answer. Your script contains several nice features that I will incorporate in my own script.
I have always been using "stage" in my script. You use "install". What is the difference?
I've never really used stage, so I might get this wrong. As I understand it, stage is designed to place the build results in the source tree. I think under a "stage" directory just under the top level. "install" is designed to place the build results in a location that generally makes sense for the operating system. In Unix, I believe it puts libraries and header files in /usr/local. In Windows it defaults to C:\boost. In either operating system, you can use "--prefix" to change this. The structure of these directories is as follows: C:\ + boost + lib + include + boost-1_43 + boost + all headers and subdirectories here I must admit I don't know if "install" make any attempts to customize the include folder, or if it just copies every header file it can find. Although I just found mpi and python folders when I know I told it no mpi or python, which suggests it just copies all header files. I don't know if "stage" goes to the trouble of creating an include directory under its output directory. If you still prefer stage, I believe it has an equivalent option to "--prefix". The same rules should apply (32 and 64-bit variants need separate stage directories, different boost versions and toolsets can share the same stage directory).