Re: [Openexr-devel] OpenEXR Windows port

[E. Scott Larsen]

> Re: [Openexr-devel] OpenEXR Windows port
> From: Dustin Graves
> Subject: Re: [Openexr-devel] OpenEXR Windows port
> Date: Fri, 24 Jan 2003 00:22:12 -1000
> User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US;rv:1.2) 
> Gecko/20021126
> Last night I did a rough port to Windows with Visual Studio 6 + Intel 5.0 
> Compiler. I was unable to use the compilers provided with both Visual Studio 
> 6.0 and .NET because they do not recognize this as legal syntax:
>
> template <class T>
> template <class S>
> inline bool
> Vec2<T>::operator != (const Vec2<S> &v) const
> ..
>
> The template followed by a template causes a syntax error. This is a bug with 
> the compiler, which I'll try reporting to Microsoft.

They already know about it and haven't shown any intention of fixing it 
in the last 4 years (at least) since they've known about it.
But I've figured out a work around, I'll start with your project files 
and stuff (_thanks_ for those!) and see what I can do, I'll get back to us.

> But I was able to compile most of the code with the Intel compiler. Everything 
> except IlmImfTest and ImathTest, because they use the drand48 family of 
> functions which are not provided with Visual Studio. The port was pretty 
> simple, requiring few changes. The most prominent changes were:

I've got a full license for it, so it'll help (shouldn't mention here 
that it is a superior compiler, so I won't).

> 1. typedef Int64 as __int64
> 2. the numeric suffix UUL is not supported, and must be changed to UL
> 3. std::min and std::max are unavailable (due to conflict with min and max in 
> windefs.h), so std::_MIN and std::_MAX are provided instead.

I always undef those after including windows.h, but your solution looks 
fine too.

> 4. Imath needed a preprocessor directive to indicate hypot should be used 
> instead of hypotf
> 5. Scoping error with variable declarations within 'for' loops, resulting in 
> multiple defines, required variables to be declared outside of the 'for' loop, 
> once per function.

again, non-spec compilers.  I've seen someone do this:
{for (int i......){
        //code
}}
so that all compilers scope the variable declaration properly.  It's 
always worked where I work. (MIPSPro, gcc, MSVC, Intel)

//Scott

> 6. windows.h must be included before gl.h
>
> I think that was pretty much it. I have put the VC6 project files, patches with 
> the changes I made for the Windows port, and precompiled binaries (libs for 
> Imath, Iex, Half, IlmImf, and an executable for exdisplay) on the web. You can 
> find them at:
>
> http://www.hawaiian.net/~dgraves/openexr/openexr.html
>
> Also, Intel has a 30 day trial period for their compiler for anyone interested. 
> You can optain it from their website. The installer will integrate it with 
> Visual Studio, allowing you to select the compiler from the Tools menu. But, 
> hopefully if the problem with templates and their template members is somehow 
> resolved the Intel compiler will be unnecessary.
>
> Dustin