Re: [Chicken-users] CMake/MinGW build 325 is using wrong (Unix) filename separat

[Brandon J. Van Every]

John Cowan wrote:

Brandon J. Van Every scripsit:

  

Using MinGW under a Cygwin shell sounds pathological, kinky, error 
prone, unlikely, and not something we should worry about supporting.  
    

On the contrary: it is a very reasonable and principled thing to do,
not different from cross-compiling Win32 code under Linux.
  

I'm not buying this one at face value.  I hear of Linuxers cross-compiling with MinGW for Windows.  Never heard of Cygwin people doing that.  You find me a large community of Cygwinners actually doing that, I'll change my tune.  I'm not interested in principles, I'm interested in what provably works in practice, replicated by thousands of people.  As always, if there's a Linux-MinGW champion or a Cygwin-MinGW champion out there, I'll work with them, but I won't take on such support burdens myself.  They're of no value to me, setting up such build environments is a real pain, and in the Linux cross-compiling case I don't have a Linux box.

There is the additional issue of what CMake supports.  I doubt they've implemented a Linux MinGW cross-compilation generator.  But, I don't know if such a generator is necessary.  I think they'd look at me funny if I told them I wanted to use a Cygwin shell to build MinGW, given that they gave me MSYS support.

The Q language <http://q-lang.sf.net>, for example, is released for
both *ix and Windows systems.  The Windows version is built under
Cygwin using MinGW to be executed on systems that have neither.
(It can also be built natively under Cygwin, of course.)
  

And how do they install their MinGW?  Do they go to www.mingw.org, or do they use the "mingw" stuff in a Cygwin FTP package?  Those are rather different cases.

How many other examples can you name?


Cheers,
Brandon Van Every