Re: Re: cvs: ZendEngine2(PHP_5_3) / zend.c zend.h zend_vm_def.h zend_vm_execute.h php-src NEWS php-src/main fopen_wrappers.c fopen_wrappers.h main.c
| From: | Marcus Boerger | Date: | Thu, 06 Mar 2008 17:18:44 +0000 |
| Subject: | Re: Re: cvs: ZendEngine2(PHP_5_3) / zend.c zend.h zend_vm_def.h zend_vm_execute.h php-src NEWS php-src/main fopen_wrappers.c fopen_wrappers.h main.c | ||
| References: | 1 2 | Groups: | php.internals |
| Request: | Send a blank email to internals+get-36006@lists.php.net to get a copy of this message | ||
Hello Gregory,
for include require we anyway use the streams stuff, don't we?
marcus
Wednesday, March 5, 2008, 9:24:39 PM, you wrote:
> Dmitry Stogov wrote:
>> dmitry Wed Mar 5 13:34:16 2008 UTC
>>
>> Modified files: (Branch: PHP_5_3)
>> /php-src NEWS
>> /php-src/main fopen_wrappers.c fopen_wrappers.h main.c
>> /ZendEngine2 zend.c zend.h zend_vm_def.h zend_vm_execute.h
>> Log:
>> Optimized require_once() and include_once() by eliminationg open() syscall on second
>> usage.
> Hi Dmitry,
> This is a very nice addition - I wonder if you might be interested in a
> patch to enable stream wrappers in include_path? There are a couple of
> ways this could be done, but the one I think would be best is checking
> for "//" immediately following ":" and then attempting to locate a
> stream wrapper, and if not found then shunt to the current
> realpath/fopen behavior.
> I'm already implementing this kind of thing already inside pecl/phar, it
> would be trivial to provide a patch and a few tests.
> The only question is whether this would be better resolved by simply
> intercepting zend_resolve_path inside pecl/phar, because include_path
> scanning may be significantly slower if we have to check for all
> possible stream wrappers and the requisite security checks for url-based
> wrappers. Checking for phar is very efficient as I've currently
> implemented it (have not yet committed, there are a few tests to be
> written/details to perfect), and doesn't open the door for
> shoot-yourself-in-the-foot security issues since all phar archives must
> be local on-disk.
> Thanks,
> Greg
Best regards,
Marcus