Re: [Simulavr-devel] Windows port of simulavrxx

[E. Weddington]

Bill wrote:

> right...a bit off topic I suppose.
>
> What I had in mind for Microsoft platforms was to support the "cygwin"
> environment. (www.cygwin.com)
Right. That's the first step. Currently simulavr builds under cygwin. 
(Remember, I put together WinAVR.)

> I don't know if it currently builds there. I know of several
> shortfalls, but the core simulator might be pretty close already.
I don't know if simulavrxx builds under cygwin or not. There may be a 
high probability that it will.

> If there are specific ideas we need to think about, them please do
> share them. Unless the volume goes way up, I don't mind direct email
> to support this.
Ideally, it would be good to eventually have a native win32 API port of 
simulavrxx, similar to what has happened with avrdude. The main reason 
why this is ideal is because it is difficult to redistribute binaries 
that are linked to the Cygwin DLL (cygwin1.dll) due to DLL version 
conflicts (I've discussed this a number of times on multiple lists.)

One of the first things to do is to identify where POSIX calls are being 
made. A couple of options then could be:
1. Isolate the POSIX calls to their own modules, then win32 modules can 
be built to provide the same functionality.
2. I recently discovered GNUlib:
Savannah page:
<http://savannah.gnu.org/projects/gnulib>
Home page:
<http://www.gnu.org/software/gnulib/>
GNUlib is a portability library which has similarities to libiberty in 
GCC. (Ted, if you're reading this, I would also suggest this for avarice 
as well.) The down side is there doesn't seem to be any functionality 
regarding sockets (which IIRC is needed for simulavr[xx] and avarice?)

Eric