Front page | perl.perl5.porters |
Postings from May 2021
From: Nicholas Clark
May 19, 2021 19:33
Message ID: 20210519193338.GY16703@etla.org
On Wed, May 19, 2021 at 04:19:38PM +0100, Steve Hay via perl5-porters wrote:
> I've tested each of your 8 test cases with each MSVC compiler that I have:
Thanks. That is really useful.
If I try to re-order that into "what we get when"
Microsoft Visual C++ 6.0 Professional x86 (VC6.0)
* // comments
* flexible array members
neither of which give us any functionality we don't already have
Microsoft Visual C++ 2003 Toolkit x86 (VC7.1)
* 64 bit types (ie unsigned long long)
Microsoft Visual C++ 2005 Express x86 (VC8.0)
* variadic macros
Microsoft Visual C++ 2013 Express x86 (VC12.0)
* mixed declarations and code
* declarations in for loops
* member structure initialisers
> These are the free versions of VC7 through VC14.2, which mainly limits
> them to being x86. I have access to paid x64 versions too if you want
> me to check, but I'm 99% sure the results would be the same.
I wouldn't bother. I'm going to assume the same.
> All of the above compilers are running fine on my Windows 10 system on
> my Windows 7 system before it. I suspect they would all run on Windows
> Vista too.
Even if they don't, can they *target* Vista?
IIRC one of the reasons "we" were keen to stick to VC6.0 was because it was
the last version that could target the C runtime DLL that was shipping on
all existing Win32 versions. (Or something close to that). In that, one
could ship an EXE compiled with it, and not have to worry about using an
installer to also ship DLLs.
(Related to MS politics about who controlled the C runtime DLLs - the
OS group, or the compiler group)
I'm probably misremembering the details, but *that* also got resolved in
the past five-ish years, didn't it? In that distribution became easier