During function calls arguments are pushed onto the stack. Now these are
backed up on yield and restored on resume. This requires memcpy'ing them,
but there doesn't seem to be any better way to do it.
Also this fixes the issue with exceptions thrown during function calls.
The missing piece is how one can find the next stack frame, which is
required for dtor'ing arguments pushed to the stack. As the generator
execute_data does not live on the stack one can't use it to figure out the
start of the next stack frame. So there must be some other method.
When no key is explicitely yielded PHP will used auto-incrementing keys
as a fallback. They behave the same as with arrays, i.e. the key is the
successor of the largest previously used integer key.
Keys are yielded using the
yield $key => $value
syntax. Currently this is implemented as a statement only and not as an
expression, because conflicts arise considering nesting and use in arrays:
yield yield $a => $b;
// could be either
yield (yield $a) => $b;
// or
yield (yield $a => $b);
Once I find some way to resolve these conflicts this should be available
as an expression too.
Also the key yielding code is rather copy-and-past-y for the value yielding
code, so that should be factored out.
If the generator is used as a coroutine it often doesn't make sense to yield
anything. In this case one can simply receive values using
$value = yield;
The yield here will simply yield NULL.
Yield now is an expression and the return value is the value passed to
$generator->send(). By default (i.e. if ->next() is called) the value is
NULL.
Unlike in Python ->send() can be run without priming the generator with a
->next() call first.
To keep things clean two new functions are introduced:
zend_clean_and_cache_symbol_table(HashTable *symbol_table)
zend_free_compiled_variables(zval ***CVs, int num)
If the generator is closed before it has finished running, it may happen
that some FREE or SWITCH_FREE opcodes haven't been executed and memory is
leaked.
This fixes it by walking the brk_cont_array and manually freeing the
variables.
For generators ZEND_RETURN directly calls ZEND_VM_RETURN(), thus passing
execution back to the caller (zend_generator_resume).
This commit also adds a check that only return; is used in generators and
not return $value;.
* pull-request/54:
Allow arbitrary expressions for empty()
This change is as per RFC https://wiki.php.net/rfc/empty_isset_exprs.
The change allows passing the result of function calls and other
expressions to the empty() language construct. This is accomplished by
simply rewriting empty(expr) to !expr.
The change does not affect the suppression of errors when using empty()
on variables. empty($undefinedVar) will continue not to throw errors.
When an expression is used inside empty() on the other hand, errors will
not be suppressed. Thus empty($undefinedVar + $somethingElse) *will*
throw a notice.
The change also does not make empty() into a real function, so using
'empty' as a callback is still not possible.
In addition to the empty() changes the commit adds nicer error messages
when isset() is used on function call results or other expressions.
This fixes the fix for bug #54547 in 32-bit machines by accepting
float comparisons in 32-bit machines as long as the integer is
not larger than the mantissa.
The Generator class now uses a zend_generator struct, so it'll be able to
store additional info.
This commit also ensures that Generator cannot be directly instantiated
and extended. The error tests are now in a separate folder from the
(yet-to-come) functional tests.
This change is as per RFC https://wiki.php.net/rfc/empty_isset_exprs.
The change allows passing the result of function calls and other
expressions to the empty() language construct. This is accomplished by
simply rewriting empty(expr) to !expr.
The change does not affect the suppression of errors when using empty()
on variables. empty($undefinedVar) will continue not to throw errors.
When an expression is used inside empty() on the other hand, errors will
not be suppressed. Thus empty($undefinedVar + $somethingElse) *will*
throw a notice.
The change also does not make empty() into a real function, so using
'empty' as a callback is still not possible.
In addition to the empty() changes the commit adds nicer error messages
when isset() is used on function call results or other expressions.
* PHP-5.4:
Fixed Bug #62005 (unexpected behavior when incrementally assigning to a member of a null object)
fix stack overflow in php_intlog10abs()
fix stack overflow in php_intlog10abs()
* PHP-5.3:
Fixed Bug #62005 (unexpected behavior when incrementally assigning to a member of a null object)
fix stack overflow in php_intlog10abs()
Conflicts:
Zend/zend_execute.c
Generate T_STRING_VARNAME only if it actually is one. This is only the case
for "${varname}" and "${varname[offset]}" so we can just add a check for
} or [ after the LABEL.
# The handling of private properties in classes is now consistent with private properties in traits.
# Perviously, privates could cause strict warnings, are were not properly merged into the class when
# the parent class had a private property of the same name. Now, we introduce it without notice,
# since it is a new and independent property, just like in normal classes.
# This problem was diagnosed while working on Bug #60536.
# The handling of private properties in classes is now consistent with private properties in traits.
# Perviously, privates could cause strict warnings, are were not properly merged into the class when
# the parent class had a private property of the same name. Now, we introduce it without notice,
# since it is a new and independent property, just like in normal classes.
# This problem was diagnosed while working on Bug #60536.
# this is a tough one, I think I should explain
# Zend use zend_object->properties_table both as zval ** and zval ***
# if a zend_object->properties is not initialized, the properties_table is zval **
# while in rebuild_object_properties, zend will store the zval ** to zend_object->properties
# then stash the zval ***(ie, zobj->properties_table[0] is zval ** now) to zobj->properties_table[0]
# so when a zend_object inherit form multi parent and these parent have a same property_info->offset
# properties, will result in a repeat zval **->zval ** transform, which will lead to a segmentfault
# *may be* this fix is not the best fix, we should not use this tricky way, and rewrite this mechanism.
# this is a tough one, I think I should explain
# Zend use zend_object->properties_table both as zval ** and zval ***
# if a zend_object->properties is not initialized, the properties_table is zval **
# while in rebuild_object_properties, zend will store the zval ** to zend_object->properties
# then stash the zval ***(ie, zobj->properties_table[0] is zval ** now) to zobj->properties_table[0]
# so when a zend_object inherit form multi parent and these parent have a same property_info->offset
# properties, will result in a repeat zval **->zval ** transform, which will lead to a segmentfault
# *may be* this fix is not the best fix, we should not use this tricky way, and rewrite this mechanism.
- aliases that are not actually matching anything are treated as errors now. This
will make sure that all methods that are expected to be in a class are actually
there, or in case a trait changed for instance, that the code breaks already
on composition
- Precedence declarations are also checked to ensure that the method
which is supposed to take precedence actually exists, however,
the other traits mentioned in the declaration are not regarded.
We are more lenient here, since this avoids unnecessary fragility.
- fixed another seamingly unrelated test which broke in the progress
but wasn't clear before either.
- aliases that are not actually matching anything are treated as errors now. This
will make sure that all methods that are expected to be in a class are actually
there, or in case a trait changed for instance, that the code breaks already
on composition
- Precedence declarations are also checked to ensure that the method
which is supposed to take precedence actually exists, however,
the other traits mentioned in the declaration are not regarded.
We are more lenient here, since this avoids unnecessary fragility.
- fixed another seamingly unrelated test which broke in the progress
but wasn't clear before either.
# Moved the freeing of overriden functions to a point after the check.
# The new check comes after the normal inheritance check to give the first check
# the opportunity to abort with a more detailed error.
# Also fixed a small type in an unrelated test.
# Moved the freeing of overriden functions to a point after the check.
# The new check comes after the normal inheritance check to give the first check
# the opportunity to abort with a more detailed error.
# Also fixed a small type in an unrelated test.
# The handling of legacy constructors defined by traits was corrected.
# They are now properly registered and used on instantiation.
# The situation for conflicting legacy and __construct constructors is
# mostly identical. If they are defined in the class, they override conflicts
# and do not collide. However, in case different styles are mixed, between
# class and trait definition, we assume a programmer's mistake and report
# a collision.
#
# BTW: +1 for all the fixed tests! `make test` is fun again.
# The handling of legacy constructors defined by traits was corrected.
# They are now properly registered and used on instantiation.
# The situation for conflicting legacy and __construct constructors is
# mostly identical. If they are defined in the class, they override conflicts
# and do not collide. However, in case different styles are mixed, between
# class and trait definition, we assume a programmer's mistake and report
# a collision.
#
# BTW: +1 for all the fixed tests! `make test` is fun again.
# The method got unconditionally deleted from the class, since it was assumed that we override it, but we did not in case of abstract methods coming from a trait. Thus, dont delete when we try to merge in an abstract method.
# The method got unconditionally deleted from the class, since it was assumed that we override it, but we did not in case of abstract methods coming from a trait. Thus, dont delete when we try to merge in an abstract method.
