Re: [boost] [regression] New runners for old compilers MSVC-7.1 and GCC 3.4

Subject: Re: [boost] [regression] New runners for old compilers MSVC-7.1 and GCC 3.4
From: Edward Diener (eldiener_at_[hidden])
Date: 2015-10-16 14:10:44

On 10/16/2015 8:22 AM, Ion Gaztañaga wrote:
> On 16/10/2015 10:43, Andrey Semashev wrote:
>> On 16.10.2015 11:18, Ion Gaztañaga wrote:
>>> Hi,
>>>
>>> The last compiler update two years ago in Boost 1.55 left MSVC-7.1 and
>>> GCC 3.3 as two of the oldest supported compilers:
>>>
>>> http://www.boost.org/users/history/version_1_55_0.html
>>
>> That's about Boost.Config supporting macros from these compilers up. It
>> doesn't mean that all libraries support these compilers.
>
> Right. I never said that, it's up to each library which compilers are
> supported. In any case, Config changes break nearly all libraries.
>
>> FWIW, I think these ancient compilers are long beyond saving. We have
>> not been testing them for years, and from the test matrix it is clear
>> that MSVC-7.1 has been broken terribly but noone complained. Noone seems
>> to care about it anymore.
>
> It's an option. Qt only supports MSVC10/GCC 4.4 and above. However
> latest POCO (1.6.1) still supports MSVC-7.1, WxWidgets supports MSVC8.
>
> I don't think MSVC is the main problem now. We have much more work with
> compilers like old DMD, Oracle, & Borland which as less conformant than
> MSVC-7.1

No one has suggested adding support for DMD or Borland in some library 
where it did not exist before, as you have done for VC++7.1. As for 
Oracle the Sun compiler 12.4 is fairly recent and even 12.2 and 12.3 
were not that long ago. The VC++7.1 compiler is from 2003, some 12 years 
ago, and Microsoft has released 6 VC++ compiler updates since then.

>
> In any case, if this triggers a discussion about new minimum
> requirements for Boost.Config, I'd happy to test other newer platforms.