[CMake] Cygwin drive paths are killing me

[Marco Atzeri]

On 15/06/2014 23:45, J Decker wrote:
>
>     Of course someone like me find some upsides, like having a
>     large number of unix tools available on windows, otherwise
>     we will have no users nor any packagers
>
>
> I guess over the years I've just ended up collecting small static
> versions of 'unix tools' in my tools folder under unix....
>
> cp, rm, mv, ls, mkdir.exe, echo.exe, touch, uname, make.exe,
>
> and some less generally useful things, wget, ssh, cvs, mtn
>
> but I think of the above cmake only needs make and other things are
> handled with cmake -E commands
> and then the compiler...  and  make (mingw32-make) comes with mingw
> anyhow; so I'm pretty sure everything can build with MinGW (not MSYS),
> cmake, and the android NDK on a clean system.

of course.

I found curious on this "android NDK" the usage of a cygwin build system
with a not cygwin compiler. I would expect of both or none, mixing
is really complicating the things.
Last time I cross compiled from windows to a Linux MIPS,
the cross compiler was a true cygwin application.

> (and 30% faster than a cygwin environment)

Likely, cygwin is handicapped by the lack of a true system fork.

> personally I can't stand how slow cygwin is; even things just built with
> cygwin run 30-50% slower than any other compiler... and it's really not
> worth it to have paths that don't really work.... and then redhat got
> ahold of it and it just went downhill from there (yes that was a long
> time ago)
>
> MinGW doesn't even need environment changes, just need the bin in PATH...

cygwin also does not need environment changes. But this is more a
discussion for another mailing list.

Regards
Marco