Python MSVC++ binaries considered evil

[Edward Diener]

I use both C++ Builder 5 and VC++ 6 pretty regularly. VC++ has retained 
some non-compliant features for a long time, with Microsoft's excuse 
being mainly that it will break old code to change. VC++ does not 
support partial specialization at all, and even the new VC++ 7.0 doesn't 
either although it is rumored that VC 7.1 with Stan Lippman at the helm 
of VC++ will. There are a number of template features in VC++ 6 that do 
not work properly. OTOH C++ Builder has problems with non-type template 
parameters.

But rather than argue one compiler against another, why don't you just 
download the free BCC compiler and test it out. BCB6 and VC7 are coming 
out pretty much together so there will be a new round of comparisons no 
doubt.

As Alex Martelli states in a further post to this thread Comeau C++ is 
far more C++ standard compliant than either VC++ or BCC. I have also the 
greatest respect for Greg Comeau and have even jokingly suggested to him 
that Borland should just pay him what he wants to get their own compiler 
up to speed. However, I don't think that Comeau C++ for Windows is free. 
  I wouldn't discourage anyone for paying for it if they were 
considering C++ standard compliance. If it is free, from just the 
viewpoint of compliance, it is easily the best for Windows.

Greg Landrum wrote:

> "Edward Diener" <eldiener at earthlink.net> wrote in message
> news:3C524424.8040601 at earthlink.net...
> 
>>It should be very possible to create a Python distribution using
>>Borland's free compiler. It has some bugs, as does VC++, but is
>>generally more compliant to the C++ standard than VC++.
>>
> 
> Just out of curiousity, do you have any references to back up this
> assertion?  I'm on the lookout for a compiler which is closer to the
> standard than VC++.