The build sequence for Appveyor is broken. Need to test how to get CGO working with Go again. Not entirely sure why CGO support is even needed.
Option 1 is to uninstall Go 1.5 and install 1.4 instead. It seems like this is needed for bootstrapping Go 1.5 anyway with CGO support.
Option 2 Is to find a build process that doesn't rely on Appveyor built-in Go and just installs a custom Go along with Mingw. Unfortunately, there isn't a tiny compiler for GCC.
Option 3 Switch to Azure with a custom build that includes mingw. This will hopefully allow for installing 32-bit OS or at least 32-bit versions of both Go and mingw to test 32-bit versions of the library. Hopefully, the tests will run faster too. This option would cost the most money.
Option 4 Might need to do caching or do this regardless of the above choices. Caching should allow for storing the mingw and go installations and install based on CI options. This should speed up the builds.
The build sequence for Appveyor is broken. Need to test how to get CGO working with Go again. Not entirely sure why CGO support is even needed.
Option 1 is to uninstall Go 1.5 and install 1.4 instead. It seems like this is needed for bootstrapping Go 1.5 anyway with CGO support.
Option 2 Is to find a build process that doesn't rely on Appveyor built-in Go and just installs a custom Go along with Mingw. Unfortunately, there isn't a tiny compiler for GCC.
Option 3 Switch to Azure with a custom build that includes mingw. This will hopefully allow for installing 32-bit OS or at least 32-bit versions of both Go and mingw to test 32-bit versions of the library. Hopefully, the tests will run faster too. This option would cost the most money.
Option 4 Might need to do caching or do this regardless of the above choices. Caching should allow for storing the mingw and go installations and install based on CI options. This should speed up the builds.