Closes GH-5353. From now on, PHP will have reflection information
about default values of parameters of internal functions.
Co-authored-by: Nikita Popov <nikita.ppv@gmail.com>
There is no reason to expect a `1` after the PID; neither the session
ID nor the memory usage are required to contain one. Actually, we just
want to verify here, that the process with the $child_pid is running,
and is a php.exe process.
Provides the last PCRE error as a human-readable message, similar
to functionality existing in other extensions, such as
json_last_error_msg().
Closes GH-5185.
We also change `Generator::throw()` to expect a `Throwable` in the
first place, and we now throw a TypeError instead of returning `false`
from `Exception::getTraceAsString()`.
There are two related changes here:
1. Also check for S_ISCHR/FILE_TYPE_CHAR when checking for pipes, so
that we detect ttys as well, which are also not seekable.
2. Always set position=-1 (i.e. ftell will return false) when a pipe
is detected. Previously position=0 was sometimes used, depending on
whether we're on Windows/Linux and whether the FD or FILE codepath
was used.
Also adds an --asan flag to run-tests.php to setup all the necessary
environment variables. Some tests are marked as skipped because they
are incompatible with asan or too slow.
I'm basing this on the DEBUG_ZTS build, which seems to give us the
most mileage.
This test is failing on AppVeyor almost all of the time, so splitting
it seems appropriate. This also allows us to rid php_cli_server_stop()
which was only used by this test case.
php_win32_signal_system_ctrl_handler() is called from a kernel thread,
so the former initialization of `vm_interrupt_flag` has no effect,
since it is defined as thread-local. This is, however, not necessary,
since the CTRL signal handling is supposed to work only for the main
thread anyway. We therefore change `vm_interrupt_flag` and the related
variables to true globals.
This also allows us to unmark the respective test case as XFAIL.
Furthermore, `vm_interrupt_flag` is declared as `zend_bool *`, so we
better treat it such.
This patch follows previous license year ranges updates. With new
approach source code files now have simplified headers with license
information without year ranges.
I thought these were redundant, because we already NULL out the
clone_obj object handler. However, it turns out that reflection is
using private __clone() to determine clonability (isCloneable) for
the case where we only have a class, rather than an object.
As such, removing these methods would be a BC break.
This reverts commit e7131a4e9f.
This reverts commit 55bd88ce0d.
Exceptions already prohibit cloning by setting clone_obj to NULL
(which is integrated with reflection). No need to additionally
define a dummy __clone() method.
The binary can be of course used on console, for whatever reasons, so
UNICODE API should be used in that case. That might however not work as
expected, if the binary is used for a service.
