typeof #1932
typeof #1932
Conversation
|
I will try to reproduce the build failure on my setup and fix it. PS: Please do not tell me where the problem is, I would prefer to find it on my own. I already prepared a Linux VM to run the tests similarly to Travis CI. |
Previously not enough memory was allocated and it was just pure luck that it worked in my initial tests.
The problem is actually extremely obvious to the experienced C developer. The allocated memory for the type variable must be freed after it was copied for returning to userland.
| @@ -2563,6 +2564,11 @@ ZEND_BEGIN_ARG_INFO_EX(arginfo_is_callable, 0, 0, 1) | |||
| ZEND_ARG_INFO(0, syntax_only) | |||
| ZEND_ARG_INFO(1, callable_name) | |||
| ZEND_END_ARG_INFO() | |||
|
|
|||
| ZEND_BEGIN_ARG_INFO_EX(arginfo_typeof, 0, 0, 1) | |||
There was a problem hiding this comment.
nit; Use ZEND_RETURN_VALUE instead of 0 in the third arg. It's better for BC and more readable.
There was a problem hiding this comment.
I don't think anybody does this in core.
There was a problem hiding this comment.
I changed it now, the readability argument directly convinced me.
| @@ -434,6 +450,135 @@ PHP_FUNCTION(is_callable) | |||
| } | |||
| /* }}} */ | |||
|
|
|||
| #define RETURN_TYPE_NAME(t) { RETURN_STRINGL(t, sizeof(t) - 1) } | |||
There was a problem hiding this comment.
minor; Create interned strings for these return values so that you're not creating a new string on every invocation.
There was a problem hiding this comment.
That would only apply to the extended mode, right? The constants are already constant and the sizeof will be replaced with the actual number anyways during compilation. Just asking to gain a better understanding of C.
Reading up on interned strings now.
More readable and better for BC; according to @sgolemons feedback.
|
Closing in favor of #1935. |
Proposal for a
gettype()successor.