Updated UPM cmake to build both python2/3 modules if the necessary
python libs are available.
* Removed BUILDPYTHON3 cmake option in favor of building both
versions
* Added cmake module to find both python2 and python3
* Added macro which can be called from per-sensor cmake files
to add dependency libraries via target_link_libraries as well
as swig_link_libraries (for java, nodejs, and python)
* Python2 is required for python documentation builds
* Cleanup of python usage throughout UPM cmakelists.
* Make find_package for Node required. Removed check for NODE_FOUND
(which should have been NODEJS_FOUND) in favor of REQUIRED.
* The cxx and .py wrapper files get generated in python2/3 (instead
of the same directory). It appears these files are identical,
however since the targets can be built in parallel it's safer to split
them out.
* Updated all cmake dependencies related to building documentation.
* Removed unused classname from cmake files
Signed-off-by: Noel Eck <noel.eck@intel.com>
Test commit for building C UPM modules.
* Added C include directory
* Added C utilities directory
* Rename C++ upm.h -> upm.hpp to make room for C upm.h
* Added upm_mixed_module_init function to src/CMakeLists.txt. This
function takes filesnames similar to upm_module_init and does a
bit of processing before calling upm_module_init.
* Added c example directory. Changed c++ example names.
* Added dfrph implemention for testing (C++ wraps C). Added mraa
to .pc requires for dfrph. Tested against stand-alone project.
Added dfrph c example.
* Update implemention of pkg-config file generation.
* Added two cmake cache variables: BUILDCPP and BUILDFTI
* Removed src from swig_add_module calls, added libname to
swig_link_libraries calls. Shrinks swig'ed binaries by ~13%.
* Added install target in upm/CMakeLists.txt to install C header,
directory. Is this where we want this?
* C FTI header directory is include/fti
Signed-off-by: Noel Eck <noel.eck@intel.com>
HD44780 fails to initialize unsuccessfully without the workaround.
Adding workaround based on HD77480 datasheet for initialization
and software reset.
Signed-off-by: Akira Tsukamoto <akira.tsukamoto@linaro.org>
Signed-off-by: Abhishek Malik <abhishek.malik@intel.com>
To make room for UPM C and C++ sensor code to coexist, all UPM
C++ headers have been renamed from h -> hpp. This commit contains
updates to documentation, includes, cmake collateral, examples, and
swig interface files.
* Renamed all cxx/cpp header files which contain the string
'copyright intel' from .h -> .hpp (if not already hpp).
* Replaced all references to .h with .hpp in documentation,
source files, cmake collateral, example code, and swig interface
files.
* Replaced cmake variable module_h with module_hpp.
* Intentionally left upm.h since this file currently does not
contain code (documentation only).
Signed-off-by: Noel Eck <noel.eck@intel.com>
Otherwise, we generate an exception. This should fix Issue #172:
https://github.com/intel-iot-devkit/upm/issues/172
Signed-off-by: Zion Orent <zorent@ics.com>
Signed-off-by: Jon Trulson <jtrulson@ics.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
Using an external code parsing library we discovered that there are
non-printable characters in some source files which break the parser.
This commit removes these characters and rewrites a type definition
which was also breaking the parser.
Signed-off-by: Mircea Bardac <mircea.bardac@intel.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
Signed-off-by: Stefan Andritoiu <stefan.andritoiu@intel.com>
Signed-off-by: Andrei Vasiliu <andrei.vasiliu@intel.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
This driver supports the Sparkfun 64x48 pixel OLED Edison block:
https://www.sparkfun.com/products/13035
It is based on an ssd1306, but with some modifications (custom COM pin
mapping and a custom column offset). It uses SPI to communicate, and
since it is an Edison Block, you don't really have any options for
different bus and pin assignments.
Signed-off-by: Jon Trulson <jtrulson@ics.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
We add a new bool isExpander to the the I2C constructor. This allows
us (in the case we are not dealing with an expander) to just
initialize the i2c context only and bail in the constructor. The
default is true to preserve backward compatibility.
Additionally, add two new virtual protected methods, data() and
command() for sending data or commands to the controller. The default
implementation in lcm1602 will work for 4bit GPIO and expander use.
In the case of jhd1313m1, they will need to be redefined in that class
as the sematics for sending data and commands are different.
All of the underlying functionality will use command() or data() as
appropriate.
Signed-off-by: Jon Trulson <jtrulson@ics.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
This driver was developed with a Sainsmart LCD Keypad Shield.
Signed-off-by: Jon Trulson <jtrulson@ics.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
In addition, move the command/data sending methods into the protected
block so that derived classes can use them if need be.
Signed-off-by: Jon Trulson <jtrulson@ics.com>
Signed-off-by: Mihai Tudor Panu <mihai.tudor.panu@intel.com>
