Subject: Re: [boost] Request for a "Policy Review" regarding 'CMakeLists.txt'
From: Robert Ramey (ramey_at_[hidden])
Date: 2016-05-18 00:21:01
On 5/17/16 4:30 PM, Steven Watanabe wrote:
> My personal opinion on the subject is that
> CMake is not supported by Boost in any way,
> and any support for it is strictly at the
> discretion of individual library authors.
Correct. But it's not an opinion. Fact is that
no one has been assigned and no one has accepted
any obligation to support CMake in Boost.
> Corollary 1: There is absolutely no guarantee
> of consistency across libraries and authors
> might put CMakeLists.txt in other places
> besides the root.
Correct - there is no one who could/would make
such a guarentee
> Corollary 2: I am entirely within my rights
> in refusing to support CMake at all (even
> if patches are provided by someone else).
I'm assuming you're speaking as a library
author. This is also a fact since no such
obligation has been stated or accepted.
I would add that as far as I know, Boost
only requires the support of boost build.
It "does not prohibit any author from
supporting any other build system if he
wants to take on the task". Personally I've
had a CMakeLists.txt file in the distribution
for several years. No one has ever complained
nor used it either for all I know. For me
this has been sufficient but if someone wants
more we can add take the quoted text above
and add it somewhere. It won't change a
a thing anyway.
Now there is the entirely separate question of
how the CMake system should be used for any
given library. CMake permits a lot - some say
too much flexibility in this area to expect that
the usages are going to be in sync. If the
advocates of CMake want to provide a page -
"Guide for Boost Authors who want to support
CMake" I think it would increase the likelyhood
that there would be wider CMake support.
Honestly I don't think there's a lot to dispute
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk