Add BaylibreAcmeInstrument, a new instrument providing support for the
Baylibre ACME board as a wrapper for the IIO interface. This class
provides better access to the ACME hardware (e.g. the ability to control
the sampling frequency) and to the retrieved samples than what the other
instrument, AcmeCapeInstrument, provides. Furthermore, it removes an
unnecessary and limiting dependency by interfacing directly with the IIO
drivers instead of relying on an intermediate script ("iio-capture") potentially
introducing unexpected bugs. Finally, it allows handling multiple probes
(the ACME can have up to 8) through an easy-to-use single instance of this
class instead of having to have an instance of AcmeCapeInstrument per channel
potentially (untested) leading to race conditions between the underlying
scripts for accessing the hardware.
This commit does not overwrite AcmeCapeInstrument as
BaylibreAcmeInstrument does not provide interface compatibility with
that class. Anyhow, we believe that anything that can be achieved with
AcmeCapeInstrument can be done with BaylibreAcmeInstrument (the
reciprocal is not true) so that BaylibreAcmeInstrument might eventually
replace AcmeCapeInstrument.
Add BaylibreAcmeInstrument documentation detailing the class interface
and the ACME instrument itself and discussing the way it works and its
potential limitations.
When installing from source devlib attempts to include the commit ID in the
version of the installed pacakge however this caused issues with package
managers like pip. PEP440 specifies that local identifiers must be in the
form `<public version identifier>[+<localversion label>]` so update the
version to conform.
https://www.python.org/dev/peps/pep-0440/#local-version-identifiers
Implement PEP396 which specifies that a package should advertise its
version via __version__ attribute.
As devlib is often used as a development version directly from source,
also add a __full_version__ attribute which appends the current commit
ID to the version.
Use the __full_version__ inside setup.py
Using the wrapt module we can improve the memoize decorator. In fact, it allows
to preserve the attributes of the memoized function, such as signature,
docstring, path to the file where the function is implemented, and so on.
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
