RE: [PHP-DEV] Patch: getters/setters syntax Implementation
Ignore this patch for now, I'll get these issues addressed and I also want to improve the
performance a bit.
Felipe, I'll check out those failing tests, I found the CODING_STANDARDS doc and will read it.
With regards to performance, I'm creating the __getHours string on each invocation. I see that
Hours has the zend_literal value. What would be the best place to cache the hash_value for future
calls?
Alternatively, what about adding a get/set function pointer to the property to avoid string
manipulation/hash calculation? Only issue I see is larger (much larger?) memory footprint for
storing two additional pointers for every class property?
-Clint
-----Original Message-----
From: Felipe Pena [mailto:[email protected]]
Sent: Sunday, December 04, 2011 10:05 AM
To: Clint M Priest
Cc: Pierre Joye; [email protected]
Subject: Re: [PHP-DEV] Patch: getters/setters syntax Implementation
2011/12/4 Felipe Pena <[email protected]>:
> Hi,
>
> 2011/12/4 Clint M Priest <[email protected]>:
>> Updated patch w/o white-space:
>> http://www.clintpriest.com/patches/accessors_v1.patch
>>
>> In the end it is a relatively simple patch. The new syntax effectively creates internal
>> functions on the object and the system looks for those functions and calls them at the appropriate
>> time.
>>
>> Example:
>> class z {
>> public $Hours {
>> public get { return $this->_Hours; }
>> protected set { $this->_Hours = $value; }
>> }
>> }
>>
>> Defines:
>> $o->__getHours();
>> $o->__setHours($value);
>>
>> Standard __get()/__set() functionality checks for the more specifically defined function
>> name and calls them. I thought this would make the most sense since it would allow us to leverage
>> the existing inheritance functionality. This comes out with respect to interfaces and traits in
>> that only errors had to be changed (for clarity) on interfaces and no changes to traits were
>> necessary to support the new functionality.
>>
>> For the automatic get/set functionality, I essentially built the function body myself
>> within zend_do_end_accessor_declaration(). One point of contention here is that internally it
>> defines a __$Hours property which would be accessible from various points. I believe the standard
>> C# get/set does not allow any access to the underlying data storage. In order to accomplish that
>> there would need to be some non-standard storage or a super-private level or something. I did not
>> explore that possibility as of yet.
>>
>> I did add a couple of convenience functions that may already be available in some other
>> form I was not aware of, such as strcatalloc or MAKE_ZNODE().
>>
>> --Clint
>>
>> -----Original Message-----
>> From: Pierre Joye [mailto:[email protected]]
>> Sent: Sunday, December 04, 2011 4:50 AM
>> To: Clint M Priest
>> Cc: [email protected]
>> Subject: Re: [PHP-DEV] Patch: getters/setters syntax Implementation
>>
>> hi Clint!
>>
>>
>> Thanks for your work so far!
>>
>> On Sun, Dec 4, 2011 at 1:33 AM, Clint M Priest <[email protected]> wrote:
>>
>>> What are the next steps to get this added to some future release?
>>
>> Let discuss the implementation and how it works, then you can move to the voting phase.
>> There is no need to hurry as the next release where this patch could go in is next year.
>>
>> Cheers,
>> --
>> Pierre
>>
>> @pierrejoye | http://blog.thepimp.net | http://www.libgd.org
>>
>> --
>> PHP Internals - PHP Runtime Development Mailing List To unsubscribe,
>> visit: http://www.php.net/unsub.php
>>
>
> I've fixed the zend_compile.c and zend_object_handlers.c to build with
> --enable-maintainer-zts. (some TSRMLS_CC missing and TSRMLS_DC usage
> instead of TSRMLS_CC) Other thing I have noticed that you have not
> followed our coding standards about brackets and comments (we don't
> use the C++ style one). And about the comments looks as the patch
> isn't finished or you just forgot the remove them?
>
> http://dpaste.com/665851/plain/
>
> --
> Regards,
> Felipe Pena
Check out also the failing tests in Zend/tests/* (including segmentation fault)
--
Regards,
Felipe Pena
Thread (23 messages)