The ctrl_handler is never destroyed. We have to destroy it at request
end so we avoid leaking it and also avoid keeping a reference to
previous request memory in a next request. The latter can result in a
crash and can be demonstrated with this script and `--repeat 2`:
```php
class Test {
public function set() {
sapi_windows_set_ctrl_handler(self::cb(...));
}
public function cb() {
}
}
$test = new Test;
$test->set();
sleep(3);
```
When you hit CTRL+C in the second request you can crash.
This patch resolves both the leak and crash by destroying the
ctrl_handler after a request.
Closes GH-18231.
Currently the PHP Development Server appends a Date header in the
response, despite already set from user code.
Added a check condition before append the header, and a test file.
Closes GH-12363.
After preloading has executed, the executor globals for class_table and
function_table are still referring to the values during preloading.
If no request happens after that then these values will remain dangling
pointers. If then the -v option on CLI or -h option (and possibly
others) on CGI is provided, there is a double free.
Fix it by nulling the pointers explicitly after preloading has finished
to fix it for all SAPIs.
Closes GH-12311.
This is supported in both the CLI and CGI modes. For CLI this required
little changes.
For CGI, the tricky part was that the options parsing happens inside the
loop. This means that options passed after the -l flag were previously
simply ignored. As we now re-enter the loop we would parse the options
again, and if they are handled but don't set the script name, then CGI
will think you want to read from standard in. To keep the same "don't
parse options" behaviour I simply wrapped the options handling inside an
if.
Closes GH-10024.
Closes GH-10710.
I grepped for php_printf cases in main/ and sapi/ and converted the
cases which clearly indicate errors to fprintf(stderr, ...), like
suggested in the linked issue.
Closes GH-11163.
Multiple tests had to be changed to escape the arguments in shell
commands. Some tests are skipped because they behave differently with
spaces in the path versus without. One notable example of this is the
hashbang test which does not work because spaces in hashbangs paths are
not supported in Linux.
Co-authored-by: Michael Voříšek <mvorisek@mvorisek.cz>
`escapeshellarg()` is pretty useless on Windows, and there is no way to
support multiple lines (i.e. line breaks). Thus, we use `proc_open()`
instead of `shell_exec()`.
We also remove some apparently superfluous empty lines from the test
expectation; that seems to match libedit behavior on Linux.
Closes GH-9474.
* PHP-8.1:
[ci skip] NEWS
Fix GH-8952: std streams can not be deliberately closed (#8953)
intl extension, build fix for icu >= 69.x release. ubrk/ucnv_safeClone had been deprecated in favor of ubrk/ucnv_clone which does not use user provided stacks but remain thread safe.
