Re: [Sbcl-devel] Windows port using MS tools

[Robert R. <ro...@rf...>]

Nikodemus Siivola wrote:
> 2009/12/23 Robert Roessler<ro...@rf...>:
>
>> So, first, is anyone interested in a port NOT based on gcc and friends?
>
> If there are people working on non-gcc toolchains that can fix issues
> as they come up, I think not being gcc-only is a perfectly valid goal.
>
> Overall, I think this is really commandable effort -- it is good to
> see Windows getting some love. :)
>
> Re. specific issues:
>
>   * gas reputedly supports Intel syntax too, so maybe we can just
> switch the syntax over?

While that appears to be the case (gas supporting Intel syntax with 
".intel_syntax"), there is more to assembler source than opcodes... ;)

... there is the whole philosophy, whether a CPP pass is available / 
expected, basic capabilities, etc.  I am really not sure that a single 
source file *can* please both gas and [the MS assembler] MASM.  I am 
thinking that since a single version of any particular assembler source 
file is desirable, NASM may be a way to go.  Like gas, it can run on and 
target all of the OSes of interest.

>   * Can you retain GNU make while kicking out gcc? If not, well,
> portable Makefiles
>     are certanly possible... just painful.

I wouldn't dream of losing GNU Make! :)  I mentioned MSYS + VS2008/VC9 
as the build environment for this project on Windows 7 x64, and good old 
GNU Make 3.81 is right there.

But I will need to fix things like the ".o" assumption and make it 
configurable.

>   * Splitting out compiler differences into separate files sounds like
> the cleaner
>     solution in most cases. Possibly many of them can be made to
> disappear entirely
>     with a sufficiently cleaver cc-gcc.c and cc-msvc.c? That said, if
> in some cases
>     a local #ifdef is much easier than other alternatives and not too
> bad -- we could
>     probably live with a few.

All right, I will see how this works. :)

Robert Roessler