fix build for mingw32

[aroig]

Otherwise is_mingw32 is set too late and configure fails with missing
dlopen().

[mgreter]

Are you sure both lines need to be moved? Some info about your system would also be helpfull. We use the same pattern in other places. So I would like to know exactly if only the AS_CASE needs to be moved.

[aroig]

For my use case only the AS_CASE needs to be moved. I was compiling sassc for msys2 on Windows using the mingw compiler.

I figured that since both lines are related, it would be better to keep them together, but I admit I gave it very little thought.

If you prefer it any other way, I'm happy to force-push a different version here.

[mgreter]

I need to further check this. My first try with msys2 lead to the opposite direction as your fix. I so far tried only libsass build. Needed to patch it for mingw64, but once it was working I moved the check as you did and then it started to fail. Indicating that the checks are actually also needed for mingw. But I need to further check if this is dependent on msys or not.

[aroig]

hmm... that's odd.

I've just cloned current sassc master and tried building with

$ autoreconf -i
$ ./configure --host=x86_64-pc-mingw32 --prefix=/mingw64
$ make

The configure fails as follows

[...]
checking dynamic linker characteristics... Win32 ld.exe
checking how to hardcode library paths into programs... immediate
checking for library containing dlopen... no
configure: error: unable to find the dlopen() function

The reason for this is clear, I think. On line 60 in configure.ac there is a test for $is_mingw32, but this variable is only defined later. So, configure does not skip the check for dlopen() as it should on mingw32. After moving those two lines, the thing builds just fine.

Anyway, current configure.ac is broken, as it uses $is_mingw32 before it is even defined. Those conditions on $is_mingw32 will always behave as in non-mingw32 systems.

[mgreter]

I'm currently testing https://github.com/sass/libsass/blob/master/configure.ac. There is another fix needed for mingw64 as it now reports as mingw64. I'm pretty sure in some earlier tests I did it also reported as mingw32, but I might be wrong. Can you check #174?

As you see I removed the if for the check on top. I doesn't seem to be needed (in fact I guess that it made the build fail once I moved the check above). But I agree that the other check must be above the actual use, so that was indeed a bug.

[mgreter]

Closed with #174. Thanks for the report @aroig