Extend test coverage for the sr_analog_to_float() conversion routine.
Pass data which is encoded in native, little and big endian format to
the routine. Pass single and double precision floating point data, and
integer data in different unit sizes and signedness. Cover scale/offset
calculations as well.
Also cover expected failed conversion for unsupported combinations of
input data format details (half/quad precision float, u64 integer).
Adjust the grouping of test cases in the sequence while we are here.
Move SI prefix, unit, and rational number related tests out of the
analog-to-float conversion group.
We pass no message, so use ck_assert instead of ck_assert_msg. This
results in an error with check 0.15:
tests/strutil.c:157:2: error: too few arguments to function '_ck_assert_failed'
157 | ck_assert_msg(saved_locale != NULL);
Cover the recently introduced inline routines which back the preprocessor
macros for endianess conversion. Add test sequences for read and write
routines for different data types of different sizes, different endianess
formats and signedness, and include those routines which increment the
read/write position.
Introduce a new tests/conv.c source file which exercises the endianess
conversion macros. It's assumed that some use cases may break their
operation, fortunately these edge cases were not seen before, or the
unreliable operation went unnoticed. This test raises awareness of the
implementation's constraints.
This is a start, the test sequence will benefit from adding some more
cases to increase coverage.
Test the ".1" and "1." cases which are assumed to be problematic on
MacOS (or may not have been supported before a recent update). Add more
tests with leading signs as well as whitespace instead of a sign.
The previous implementation only echoed the p/q conversion results _if_
the return code signalled success but the result was unexpected. Although
the errno value for failed conversion attempts (non-zero return codes)
is not too helpful, seeing which text input failed the test is desirable.
Values like '-0.1' would be parsed as being positive, as the integral
type does not discern +0 and -0. Also allow values without leading
integral value, to match behaviour of strtod/sr_atof.
The version text length check fails for git setups that use more digits
in abbreviated hashes, as is recommended by e.g. the Linux kernel project.
Raise the upper limit for acceptable version strings, and add comments
on how the limits were determined. The test still might fail in setups
of slightly different configuration, but now it's easier to see why the
test failed, and how to adjust the test.
Signed-off-by: Gerhard Sittig <gerhard.sittig@gmx.net>
The method accepts strings with numbers in scientific or normal notation,
e.g. -1.25 or 3.37e-6. The numeric range is limited by the sr_rational
range, i.e +-9.2e18, resolution is ~19 digits.
Since Autoconf places some important feature flags only into the
configuration header, it is necessary to include it globally to
guarantee a consistent build.
This requires sr_hw_cleanup_all() and sanity_check_all_drivers()
to also take a context.
The (runtime) generation of the driver list now happens in sr_init()
and sr_driver_list() always returns that pre-generated list. This fixes
a segfault when (correctly) invoking multiple sr_init() and sr_exit()
calls with different contexts (caught by the unit tests).
This fixes bug #565.
Gerhard Sittig
Jiri Slaby
Uwe Hermann
Frank Stettner
Stefan Brüns
Soeren Apel
Aurelien Jacobs
Daniel Elstner