Actually, there is no negative zero at all. We obey Postel's law, and
still accept negative zeroes, but we store them as positive zeroes
after the conversion from string, i.e. we normalize before further
processing.
Instead of writing warning messages to `stderr`, we employ PHP's error
handling to raise `E_WARNING` even for the single case where
`bc_rt_error()` has been called, since that did not actually error out.
We choose to call `php_error_docref()` directly in libbcmath, since
there is no upstream, and since other PHP core functionality is already
used in our bundled libbcmath. Accordingly, we remove `rt.c` so it will
not be accidentally used in the future.
Besides adapting a few existing tests, we add new tests so that the
warnings are tested at least once. We also get rid of the Windows
specific tests, since the warning behavior is now supposed to be
platform-agnostic.
Since `bcpowmod()` does not support non-integral operands, we have to
truncate these in addition to emitting a respective warning. We also
have to work with the truncated values in the following.
We recognize that the division by one to enforce the truncation is
actually overkill, but we stick with it for now, and shall tackle the
issue for PHP 7.3.
We change `bcmul()` and `bcpow()` so that the result has exactly the
requested scale (i.e. decimal places) to make them consistent with the
other BCMath functions. This also changes our stance regarding bug
#52748, which had been classified as documentation problem.
We do not manipulate the numbers themselves (anymore), but rather
introduce `bc_num2str_ex()` which accepts a scale parameter that
overrides the scale of the number by omitting extraneous decimals and
adding zeros, respectively. This also allows us to get rid of
`split_bc_num()`, which fixes bug #75164 as well.
The `n_next` member of struct `bc_struct` is unused; the only code that
would use this member has been disabled since libbcmath has been
bundled nearly 17 years ago. Apparently, `n_next` has been designed to
build a linked list of free numbers, but it is doubtful whether that
would be an improvement over relying on the efficiency of the ZendMM,
so we remove the remaining support altogether.
The ABI break probably affects nobody, but would be okay for a new
minor version according to our release process, anyway.
configure.ac was introduced in 2001 with automake-1.15 and autoconf-2.50
to replace the file named configure.in.
Autotools is preparing to remove configure.in in Automake 2.0.
All new software should be using configure.ac.
This also fixes Bug #69770 where extensions are creating configure.in
Signed-off-by: Brian Evans <grknight@gentoo.org>
