Re: [Boost-build] [boost] [release] 1.64.0 Delayed because of Microsoft (version numbers)

Subject: Re: [Boost-build] [boost] [release] 1.64.0 Delayed because of Microsoft (version numbers)
From: Beman Dawes (bdawes_at_[hidden])
Date: 2017-03-20 13:05:54

On Sun, Mar 19, 2017 at 11:17 AM, Daniel James via Boost <
boost_at_[hidden]> wrote:

> On 19 March 2017 at 13:10, Tom Kent via Boost <boost_at_[hidden]>
> wrote:
> >
> > Option 1 - 14.10 Use microsoft toolset version based on cl.exe version
> -5.
> > This is the official version of the c++ toolset that microsoft has been
> > pushing (somewhere). The new $(VCToolsVersion) macro is "14.10.25017",
> this
> > macro is not available in previous versions of visual studio.
> >
> > build bootstrap would use bootstrap.bat vc1410
>
> [snip]
>
> > Option 2 - 14.1 Use the abbreviated toolset version that microsoft uses
> for
> > their toolset version. The $(PlatformToolsetVersion) macro is "141". In
> > VS2015 this was "140".
>
> [snip]
>
> > I *think* that is all the reasonable options. Let the consensus form!
>
> I disagree, I think we should limit ourselves to values that Microsoft
> actually uses. As far as I'm aware, they don't use 1410, and they
> don't use 14.1. There are already too many ways of writing the version
> number, adding more will just add to the confusion.
>

+1

>
> We should also make as few assumptions about the correspondence
> between different version numbers as possible. We shouldn't assume
> that the toolset version is the compiler minus 5. Similarly the idea
> that there's a fixed correspondence between whatever 141 and 14.10
> represents, we should try to deal with the possibility 142 - 14.10,
> 141 - 14.11, 142 - 14.11 and 237 - 18.65, or whatever Microsoft does
> 10 years down the line.
>

+1

--Beman