[CMake] CMakeifying Boost

[Brandon J. Van Every]

Rodrigo Madera wrote:
> Let's hope that Boost gets CMaked then.
>   
If you mean, will 1 author homebrew a Boost CMake for his own purposes, 
sure that could happen.  It seems it's happened several times already in 
various guises, and various people don't think that much is difficult.  
However, making a build, and making a build that other people can 
reproduce, are 2 different things.  "Works On My Box" is an epithet at 
Microsoft, for instance.  I don't place the odds of a reproducible build 
as all that high, going by my experiences with 2.2.3 + the empirical 
evidence that people are inventing CMake Boost builds independently 
without ever building upon each other's work.  If someone was really 
serious about reproducible builds, they'd get together 3 like-minded 
people, produce a proof-of-concept, then start talking to the Boost guys 
about getting it into a /contrib directory.  The Boost guys won't be 
interested in replacing their current build system right now, that's a 
non-starter.  But they might acquiesce to an alternate, unsupported 
build strategy if 3 other people already have proof-of-concept and are 
shouldering the work.

Or, they might look at it as contaminating the perceived reliability of 
Boost.  That's an argument to anticipate.  A dashboard would address 
that argument.

I realize that externally reproducible builds and politics are not the 
original poster's concern.


Cheers,
Brandon Van Every