We're currently splitting up large writes into 8K size chunks, which
adversely affects I/O performance in some cases. Splitting up writes
doesn't make a lot of sense, as we already must have a backing buffer,
so there is no memory/performance tradeoff to be made here.
This change disables the write chunking at the stream layer, but
retains the current retry loop for partial writes. In particular
network writes will typically only write part of the data for large
writes, so we need to keep the retry loop to preserve backwards
compatibility.
If issues due to this change turn up, chunking should be reintroduced
at lower levels where it is needed to avoid issues for specific streams,
rather than unnecessarily enforcing it for all streams.
Tyson Andre [Fri, 25 Oct 2019 23:57:39 +0000 (19:57 -0400)]
Optimize creation of empty arrays in json_decode
Use the shared empty array from ZVAL_EMPTY_ARRAY
For code that created an 10 arrays of 100000 empty arrays
(has the same result with `$assoc=true` and `{}`)
- This is the worst-case comparison, but I'd expect 0-length arrays to be fairly
common in regular data for json_decode
- The parser implementation was using function pointers so that third party
extension developers could reuse the json parser for their own
data structures, etc. (I think).
This PR is meant to let those third party extensions continue working
without changes.
Before this patch: In 0.126 seconds: added 97.99 MiB
After this patch: In 0.096 seconds: added 41.99 MiB
```php
<?php
$json = '[' . str_repeat('[],', 100000) . "null]";
$start_memory = memory_get_usage();
$start_time = microtime(true);
$result = [];
for ($i = 0; $i < 10; $i++) {
$result[] = json_decode($json);
}
$end_memory = memory_get_usage();
$end_time = microtime(true);
// Before this patch: In 0.126 seconds: added 97.99 MiB
// After this patch: In 0.096 seconds: added 41.99 MiB
printf("In %.3f seconds: added %.2f MiB\n", $end_time - $start_time, ($end_memory - $start_memory)/1000000);
// For objects
$json = '[' . str_repeat('{},', 100000) . "null]";
$start_memory = memory_get_usage();
$start_time = microtime(true);
for ($i = 0; $i < 10; $i++) {
$result[] = json_decode($json, true);
}
$end_memory = memory_get_usage();
$end_time = microtime(true);
// Before this patch: In 0.126 seconds: added 97.99 MiB
// After this patch: In 0.096 seconds: added 41.99 MiB
printf("In %.3f seconds: added %.2f MiB (objects decoded as arrays) \n", $end_time - $start_time, ($end_memory - $start_memory)/1000000);
```
Nikita Popov [Mon, 7 Oct 2019 11:04:06 +0000 (13:04 +0200)]
Warn on strtr(["" => "x"])
Previously:
* If only ["" => "x"] was present, the original string was returned
without warning.
* If both ["" => "x"] and at least one more element was present,
false was returned without warning.
New behavior:
* Ignore "" keys in the replacement array (and perform any remaining
replacement).
* Throw a warning indicating that an empty string replacement has
been ignored.
Tyson Andre [Tue, 29 Oct 2019 23:48:28 +0000 (19:48 -0400)]
Update documentation/comment for GH-4860
Fix folding for the new helper method.
Clarify comment in UPGRADING:
The performance on associative arrays would also improve,
as long as no offsets were unset (no gaps).
Packed arrays can have gaps.
Nikita Popov [Wed, 13 Feb 2019 10:37:32 +0000 (11:37 +0100)]
Don't check $this existence in object opcodes
We are now guaranteed that $this always exists inside methods, as
well as insides closures (if they use $this at all).
This removes checks for $this existence from the individual object
opcodes. Instead ZEND_FETCH_THIS is used in the cases where $this
is not guaranteed to exist, which is mainly the pseudo-main scope.
Allow to call XMLReader::open() and ::XML() statically
The implementation of `XMLReader::open()` and `XMLReader::XML()` still
supports calling the methods statically and non-statically. However,
as of PHP 8.0.0, calling these methods statically is not allowed,
because they are not declared as static methods. Since we consider it
to be cleaner to call these methods statically, but had deprecated to
call them statically, we properly support both variants.
We implement support for static and non-static calls by overloading, so
that non-static calls have access to the `$this` pointer.
Implement #78270: Support __vectorcall convention with FFI
To work around the limitation of the current rudimentary vectorcall
support in our patched libffi, we forbid yet unsupported declarations,
i.e. float/double parameters at certain positions (SIMD vector types
and HVA types are not supported anyway).
When getting the properties of a DatePeriod instance we have to retain
the proper classes, and when restoring a DatePeriod instance we have to
cater to DateTimeImmutable instances as well.
Elevate warnings to Error Exceptions in ext/bcmath
`bcdiv()` and `bcmod()` throw DivisionByZeroError if the divisor is 0,
which matches the behavior of the `/` and `%` operators, and `bcsqrt()`
throws ValueError for negative operands.
Tyson Andre [Fri, 25 Oct 2019 13:31:06 +0000 (09:31 -0400)]
Optimize array_slice for packed arrays with large offsets
If the offset is 100000, and there are no gaps in the packed/unpacked array,
then advance the pointer once by 100000,
instead of looping and skipping 100000 times.
Add a new test of array_slice handling unset offsets.
Nikita Popov [Sun, 27 Oct 2019 08:33:46 +0000 (09:33 +0100)]
Try one more FD in ext/standard/tests/file/php_fd_wrapper_04.phpt
For some reason FD 120 seems to exist on macos quite often, while
FD 12 did not... Let's try an even larger number, otherwise we
should just drop this test.
Nikita Popov [Thu, 24 Oct 2019 14:36:25 +0000 (16:36 +0200)]
Fix bug #78226: Don't call __set() on uninitialized typed properties
Assigning to an uninitialized typed property will no longer trigger
a call to __set(). However, calls to __set() are still triggered if
the property is explicitly unset().
This gives us both the behavior people generally expect, and still
allows ORMs to do lazy initialization by unsetting properties.
For PHP 8, we should fine a way to forbid unsetting of declared
properties entirely, and provide a different way to achieve lazy
initialization.
Nikita Popov [Fri, 25 Oct 2019 09:24:32 +0000 (11:24 +0200)]
Check class linking in VERIFY_RETURN_TYPE optimization
instanceof_function() requires linked classes. I'm not reusing
unlinked_instanceof() here, because it performs class loading,
which wouldn't be right here, I think.
Nikita Popov [Thu, 24 Oct 2019 16:11:41 +0000 (18:11 +0200)]
Remove recursive check from instanceof_interface
Parent interfaces are copied into the interface list during
inheritance, so there's no need to perform a recursive check.
Only exception are instanceof checks performed during inheritance
itself. However, we already have unlinked_instanceof for this
purpose, it just needs to be taught to handle this case.
Nikita Popov [Thu, 24 Oct 2019 15:47:35 +0000 (17:47 +0200)]
Optimize instanceof_class/interface
instanceof_class does not need to check for a NULL pointer in the
first iteration -- passing NULL to this function is illegal.
instanceof_interface does not need to use instanceof_class(), it
only has to check whether the CEs match exactly. There is no way
for an interface to appear inside "parent", it will always be in
"interfaces" only.
projects / php / log