windows use conda git

.jenkins/jenkins_buildbot_windows.bat

 REM CUDNN PATH
 set CUDNNPATH=C:\lib\cuda
 
-REM Set conda python, cuda, cmake path
-set PATH=%PATH%;C:\ProgramData\Miniconda2;%CUDNNPATH%\bin;C:\Program Files\CMake\bin
+REM Set conda python, cudnn, cmake path
+set PATH=%PATH%;C:\ProgramData\Miniconda2;C:\ProgramData\Miniconda2\Library\usr\bin;C:\ProgramData\Miniconda2\Library\bin;C:\ProgramData\Miniconda2\Scripts
+set PATH=%PATH%;%CUDNNPATH%\bin;C:\Program Files\CMake\bin
 
 set BUILDBOT_DIR=%WORKSPACE%\nightly_build
 set COMPILEDIR=C:\Jenkins\theano_cache\buildbot_windows
@@ -38,7 +39,7 @@ cmake --build . --target install
 cd ..\..
 
 REM Set conda gcc path
-set PATH=%PATH%;C:\ProgramData\Miniconda2;C:\ProgramData\Miniconda2\Library\mingw-w64\bin;C:\ProgramData\Miniconda2\Library\usr\bin;C:\ProgramData\Miniconda2\Library\bin;C:\ProgramData\Miniconda2\Scripts
+set PATH=%PATH%;C:\ProgramData\Miniconda2\Library\mingw-w64\bin
 
 REM Build the pygpu modules
 cd libgpuarray

.jenkins/jenkins_pr_win.bat

 REM CUDNN PATH
 set CUDNNPATH=C:\lib\cuda
 
-REM Set conda python, cuda, cmake path
-set PATH=%PATH%;C:\ProgramData\Miniconda2;%CUDNNPATH%\bin;C:\Program Files\CMake\bin
+REM Set conda python, cudnn, cmake path
+set PATH=%PATH%;C:\ProgramData\Miniconda2;C:\ProgramData\Miniconda2\Library\usr\bin;C:\ProgramData\Miniconda2\Library\bin;C:\ProgramData\Miniconda2\Scripts
+set PATH=%PATH%;%CUDNNPATH%\bin;C:\Program Files\CMake\bin
 
 REM Set cache dir and copy from master
 set COMPILEDIR=%WORKSPACE%\cache
@@ -34,7 +35,7 @@ cmake --build . --target install
 cd ..\..
 
 REM Add conda gcc toolchain path
-set PATH=%PATH%;C:\ProgramData\Miniconda2\Library\mingw-w64\bin;C:\ProgramData\Miniconda2\Library\usr\bin;C:\ProgramData\Miniconda2\Library\bin;C:\ProgramData\Miniconda2\Scripts
+set PATH=%PATH%;C:\ProgramData\Miniconda2\Library\mingw-w64\bin
 
 REM Build the pygpu modules
 cd libgpuarray

doc/install_windows.txt

@@ -24,12 +24,13 @@ Install requirements and optional packages
 
 .. code-block:: bash
 
-    conda install numpy scipy mkl-service libpython <m2w64-toolchain> <nose> <sphinx> <pydot-ng>
+    conda install numpy scipy mkl-service libpython <m2w64-toolchain> <nose> <sphinx> <pydot-ng> <git>
 
 .. note::
 
     * Arguments between <...> are optional.
     * ``m2w64-toolchain`` package provides a fully-compatible version of GCC and is then highly recommended.
+    * ``git`` package installs git source control through conda, which is required for the development versions of Theano and libgpuarray
 
 Package ``parameterized`` is also optional but may be required for unit testing. It is available via ``pip``.