I forgot to update the device and parser initialization functions to
store the context pointer into the objects. As a result, the internal
context pointers were always NULL.
The public api is changed to require a context object for all
operations. Because other library objects store the context pointer
internally, only the constructor functions need an explicit context
object as a parameter.
Adding the "dc_" namespace prefix (which is of course an abbreviation
for libdivecomputer) should avoid conflicts with other libraries. For
the time being, only the high-level device and parser layers are
changed.
The public header files are moved to a new subdirectory, to separate
the definition of the public interface from the actual implementation.
Using an identical directory layout as the final installation has the
advantage that the example code can be build outside the project tree
without any modifications to the #include statements.
Some Veo devices never respond to the initialization command, but have
no problem to continue the communication. Therefore a timeout with no
data received is ignored. If there happens to be a real problem, it will
be catched when sending one of the other commands afterwards.
Not all devices appear to send the last zero byte, resulting in a
timeout. If a device does send this zero byte, it is automatically
flushed when sending the next command.
To be able to cancel an operation, an application should register a
callback function that returns a non-zero value whenever the active
operaton should be cancelled. A backend can invoke this callback function
to query the application for a pending cancellation request.
Trying to send the version command immediately after the initialization
of the data cable doesn't work very well. Adding a small delay before
sending the version commands turns out to be much more reliable.
A helper function is added to simplify implementing the devic_dump()
function on top of the device_read() function, and enable progress
events automatically.
Using a resizable memory buffer allows to allocate the right amount of
memory inside the backend, avoiding having to know the required buffer
size in advance.
The logbook ringbuffer is now considered empty if any of the pointers is
outside the valid ringbuffer area. Compared to checking only against a
special empty value, this approach makes the code more robust against
invalid pointers.
Introducing a common base class allows to share more code between the
backends. Sharing the fingerprint data eliminates the need to pass it
with a function parameter.
