Pointer ABI for Windows (#16)

Windows ABI expects the return value to be passed by reference as a first argument. After a bit of experimentation we've settled on using the same Tuple{VecElement} trick as on unix (to ensure 128-bit alignment) and then defining ccall macro that handles the conversion correctly.

A few other small fixes:
- Use correct limb size for MPFR
- Use correct quadoplib name for Win32

I've also had to disable 0.7 support as it doesn't like the splatting syntax we need to use for sprint, but this doesn't seem like a big deal.

Author: RalphAS

.appveyor.yml

@@ -1,7 +1,7 @@
 environment:
   matrix:
-  - julia_version: 0.7
-  - julia_version: 1
+  - julia_version: 1.0
+  - julia_version: 1.1
   - julia_version: nightly
 
 platform:
@@ -35,3 +35,6 @@ build_script:
 test_script:
   - echo "%JL_TEST_SCRIPT%"
   - C:\julia\bin\julia -e "%JL_TEST_SCRIPT%"
+
+#on_finish:
+#  - ps: $blockRdp = $true; iex ((new-object net.webclient).DownloadString('https://raw.githubusercontent.com/appveyor/ci/master/scripts/enable-rdp.ps1'))

.travis.yml

@@ -4,8 +4,8 @@ os:
   - linux
   - osx
 julia:
-  - 0.7
   - 1.0
+  - 1.1
   - nightly
 notifications:
   email: false

Project.toml

@@ -0,0 +1,13 @@
+name = "Quadmath"
+uuid = "be4d8f0f-7fa4-5f49-b795-2f01399ab2dd"
+
+[deps]
+Requires = "ae029012-a4dd-5104-9daa-d747884805df"
+
+[extras]
+SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["Test", "SpecialFunctions"]
+

README.md

@@ -7,11 +7,10 @@ This is a Julia interface to libquadmath, providing a `Float128` type correspond
 
 ## Support
 
-Quadmath currently works on x86_64 Linux and macOS.
+Quadmath currently works on x86_64 Linux, macOS, and Windows.
 
 - It may require a new-ish version of gcc which supports `__float128` type.
 - It has not been tested on 32 bit Linux.
-- I have not had any luck getting it to work on Windows. It's probably something to do with the calling convention: if someone figures it out I would be very grateful.
 - It does not work on ARM due to the lack of libquadmath support for that platform.
 
 ## Installation

REQUIRE

@@ -1,2 +1,2 @@
-julia 0.7
+julia 1.0
 Requires