Re: svn commit: r612428 - /stdcxx/branches/4.2.x/include/rw/_config-msvcrt.h

[Martin Sebor]

Farid Zaripov wrote:

-----Original Message-----
From: Martin Sebor [EMAIL PROTECTED] On Behalf Of Martin Sebor
Sent: Wednesday, January 16, 2008 6:17 PM
To: stdcxx-dev@incubator.apache.org

Subject: Re: svn commit: r612428 - /stdcxx/branches/4.2.x/include/rw/_config-msvcrt.h

[EMAIL PROTECTED] wrote:

Author: faridz
Date: Wed Jan 16 04:04:48 2008
New Revision: 612428

URL: http://svn.apache.org/viewvc?rev=612428&view=rev
Log:
2008-01-16  Farid Zaripov  <[EMAIL PROTECTED]>

* include/rw/_config-msvcrt.h: Don't #define

_RWSTD_NO_LONG_LONG as __int64

        if long long type is supported by compiler.

IIRC, this was hardcoded to work around an MSVC 7 bug that rejected long long and unsigned long long overloads. I don't have access to 7.0 at the moment to check but if you do and can confirm this we might want to make the change conditional on 7.1 and better.

  I haven't access to MSVC 7.0 anymore since I'm working on 64-bit
Windows XP.
I can install MSVC 7.0 on the 32-bit Windows XP on VMware virtual
machine to
verify this change, but instead I propose to dismiss supporting the MSVC
7.0
as we already done with MSVC 6.0. Actually MSVC 7.0 is closer to MSVC
6.0
than to MSVC 7.1 and the stdcxx source has many workarounds for MSVC
7.0,
which could be removed. I think that supporting of the 3 latest versions
of the
MSVC is enough :)

We can't drop support for a compiler in 4.2.1 but getting rid of
it in 4.3 sounds reasonable to me. Do you want to start a formal
vote?

Martin