Commits
-------
8f46795 tests for HeaderBag
Discussion
----------
tests for HeaderBag
Hi,
This patch adds 100% tests coverage for HeaderBag.
Best regards,
Michal
Commits
-------
622102e [Process] change all documentation type to callable
Discussion
----------
[Process] change all documentation type to callable
Update documentation per #5456 instead of mixing in with restart() pull request.
Commits
-------
690e28e Convert parameter name to lowercase when removing an element from ParameterBag
Discussion
----------
[DependencyInjection] Convert parameter name to lowercase when removing an element from ParameterBag
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: #5477
License of the code: MIT
---------------------------------------------------------------------------
by sstok at 2012-09-10T08:42:30Z
ðŸ‘
---------------------------------------------------------------------------
by fabpot at 2012-09-10T09:28:16Z
Shouldn't it be done on the 2.0 branch?
---------------------------------------------------------------------------
by stloyd at 2012-09-10T09:29:35Z
@fabpot AFAIK `2.0` don't have `ParameterBag#remove()` method
---------------------------------------------------------------------------
by aderuwe at 2012-09-10T09:29:52Z
The remove method is not present in 2.0.
On Mon, Sep 10, 2012 at 11:28 AM, Fabien Potencier <notifications@github.com
> wrote:
> Shouldn't it be done on the 2.0 branch?
>
> —
> Reply to this email directly or view it on GitHub<https://github.com/symfony/symfony/pull/5478#issuecomment-8418300>.
>
>
Commits
-------
be28e56 [Routing] disallow numeric named variables in pattern
Discussion
----------
[Routing] compile check for numeric named variables in pattern
Because PHP raises an error for such subpatterns in PCRE and thus would break matching, e.g. this is not allowed as regex `(?<123>.+)`.
So add a compile time check for a non-working pattern like '/{123}'.
---------------------------------------------------------------------------
by sstok at 2012-09-06T08:31:42Z
Strangely enough Regex buddy gives no warning or error with the pattern.
Is the name all numeric invalid or just the beginning?
1e4 and 0xFF would be perfectly valid but returns true with is_nummeric()
---------------------------------------------------------------------------
by Tobion at 2012-09-06T08:59:07Z
Any numeric is not valid. I guess this limitation is unique to PHP's binding to PCRE.
I think it's because the returned matches array of of preg_match contains both the subpattern as integer index and as named variable. So having a numeric named variable would conflict as `array['1'] === array[1]`.
Commits
-------
7c5cfeb [Routing] added test why #5238 is not that easy
Discussion
----------
[Routing] added test why #5238 is not that easy
This just adds a test that wasn't covered yet and shows why #5238 is not that easy as I thought at first.
Commits
-------
0f86a33 micro-optim: replace for with foreach
4efb9fe [Form] Remove unneeded FormUtil constants
Discussion
----------
[Form] Remove FormUtil constants
The constants are not useful from outside the class as the $pluralMap is also private. So no need to expose these veriables in the API when they cannot be used in any way. Unfortunately there are not private constants, so use private static. Then I realized the variables can be removed altogether, as they are only used once anyway and the index meaning is already documented in pluralMap.
---------------------------------------------------------------------------
by empire at 2012-08-18T12:58:22Z
FormUtils is abstract class, and maybe subclass (in future) will use this constants, I think changing access modifier to `protected` is better option.
---------------------------------------------------------------------------
by Tobion at 2012-08-18T12:59:40Z
They cannot, as pluralMap is private...
---------------------------------------------------------------------------
by Partugal at 2012-08-18T14:11:17Z
extract self::$pluralMap into local variable add small speed up
4.5499801635742 ms vs 5.7430267333984 ms on 100 iterations
---------------------------------------------------------------------------
by Tobion at 2012-08-18T14:16:47Z
This is not about performance.
---------------------------------------------------------------------------
by Partugal at 2012-08-18T14:21:11Z
yes but adds
your changes vs current is
5.7430267333984 ms vs 6.4971446990967 ms on 100 iterations
---------------------------------------------------------------------------
by Tobion at 2012-08-18T14:29:48Z
How about `$map =& self::$pluralMap[$i]`?
---------------------------------------------------------------------------
by Partugal at 2012-08-18T14:59:57Z
I mean https://gist.github.com/3387253
---------------------------------------------------------------------------
by Partugal at 2012-08-18T15:01:45Z
foreach is event faster :)
(4.1971206665039 ms on my hw)
---------------------------------------------------------------------------
by Tobion at 2012-08-18T15:04:51Z
I see.
---------------------------------------------------------------------------
by Partugal at 2012-08-18T15:06:41Z
in first comment i mean code like this:
```php
$pluralMap = self::$pluralMap; // do this because access to static property is to slow
```
on my machine & is slower `$map =& $pluralMap[$i]` vs `$map = $pluralMap[$i]`
5.0 vs 4.8 ms
imho & not needed in read only code
---------------------------------------------------------------------------
by Tobion at 2012-08-18T15:15:03Z
Well, you'd need to benchmark memory too. `=&` should reduce memory primarily in this case.
---------------------------------------------------------------------------
by Partugal at 2012-08-18T15:19:35Z
```php
echo memory_get_usage() . "\n"; // 711536
$a = array_fill(5, 6000, 'banana');
echo memory_get_usage() . "\n"; // 1497632
$b = $a;
echo memory_get_usage() . "\n"; // 1497760
$b[1] = 2;
echo memory_get_usage() . "\n"; // 2283832
```
1497760 - 1497632 = 128 it is size for variable structure not for its value:
```php
echo memory_get_usage() . "\n"; // 711536
$a = 1;
echo memory_get_usage() . "\n"; // 1497632
$b = &$a;
echo memory_get_usage() . "\n"; // 1497760
```
---------------------------------------------------------------------------
by Seldaek at 2012-08-18T17:52:32Z
@Tobion http://schlueters.de/blog/archives/125-Do-not-use-PHP-references.html - search for "copy-on-write" if you don't want to read it all.
---------------------------------------------------------------------------
by Tobion at 2012-08-18T19:37:44Z
Yeah I know about copy on write. I thought there might be a difference what you assign a sub-element of an array to a variable. But apparently not.
Interestingly `$a =& $b` takes a little more memory than `$a = $b` according to `memory_get_usage ()` but not when using `memory_get_usage (true)`.
---------------------------------------------------------------------------
by bschussek at 2012-08-30T08:15:01Z
I don't like the removal of the constants. They introduce meaning into the integers and improve code clarity. The rest looks good.
---------------------------------------------------------------------------
by Tobion at 2012-08-30T13:18:19Z
My opinion of the constants:
- They are part of the public API (as const are alwalys public) but cannot be used at all, as everything else is private...
- They are each only used once.
- The meaning of the indices is already documented in `$pluralMap`
- They are not used when building `$pluralMap` so they dont imprivate code clarity and consistence either. But doing so would on the other hand make it probably more ugly. So removing them is IMO best solution.
---------------------------------------------------------------------------
by bschussek at 2012-08-30T15:21:03Z
If you really need to remove the constants, then please comment the code where they are used accordingly.
---------------------------------------------------------------------------
by Tobion at 2012-08-31T00:58:51Z
I dont see what I should comment to make it more understandable, as the the map is already assigned to a named variable like `$suffixLength = $map[1];`.
---------------------------------------------------------------------------
by bschussek at 2012-08-31T09:12:18Z
> I dont see what I should comment to make it more understandable, as the the map is already assigned to a named variable
`$map[2]` and `$map[3]` is not self-describing.
---------------------------------------------------------------------------
by Tobion at 2012-08-31T17:23:15Z
@bschussek Done.
---------------------------------------------------------------------------
by bschussek at 2012-08-31T22:13:41Z
Could you please squash your commits?
Commits
-------
1b5ad17 Revert "Removed MySQL-exclusive usage of unsigned integer from table creation"
Discussion
----------
[Security][DBAL] Revert MySQL unsigned removal
Revert "Removed MySQL-exclusive usage of unsigned integer from table creation"
This reverts commit 57694aaa94.
The problem is underlying in Doctrine DBAL change tracking and should
either be fixed or ignored there.
I opened a ticket on Doctrine Jira http://doctrine-project.org/jira/browse/DBAL-322
---------------------------------------------------------------------------
by fabpot at 2012-08-14T06:40:47Z
I will merge this PR after we have a release of DBAL that includes the fix for DBAL-322.
---------------------------------------------------------------------------
by acasademont at 2012-08-20T08:01:48Z
This was already fixed 2 weeks ago in doctrine/dbal#183 so i guess this can be closed
---------------------------------------------------------------------------
by acasademont at 2012-08-20T08:02:06Z
merged i mean
Commits
-------
3036b00 JsonResponseTest
Discussion
----------
JsonResponseTest
Hi,
This patch adds some tests for JsonResponse.
Best regards,
Michal
---------------------------------------------------------------------------
by eventhorizonpl at 2012-09-01T07:09:12Z
Done. Thanks for the review!
Commits
-------
8a3c8c9 load test
Discussion
----------
load test
Hi,
This patch add test that covers this situation
public static function load($classes, $cacheDir, $name, $autoReload, $adaptive = false, $extension = '.php')
{
// each $name can only be loaded once per PHP process
if (isset(self::$loaded[$name])) {
return;
}
Best regards,
Michal
Commits
-------
0af8778 Response tests
Discussion
----------
Response tests
Hi,
This patch adds some tests to ResponseTest.
Best regards,
Michal
---------------------------------------------------------------------------
by eventhorizonpl at 2012-09-01T09:45:16Z
Fixed, thanks for the review.
---------------------------------------------------------------------------
by eventhorizonpl at 2012-09-02T19:39:26Z
CS fixed. Thanks for the review :)
Commits
-------
c74d9a9 ResponseHeaderBag tests
Discussion
----------
ResponseHeaderBag tests
Hi,
This patch adds some ResponseHeaderBag tests. Now ResponseHeaderBag got 100% test coverage :)
Best regards,
Michal
Commits
-------
b89d4ee StreamedResponseTest
Discussion
----------
StreamedResponseTest
Hi,
This patch adds one test to StreamedResponseTest and fixes another. StreamedResponse has 100% test coverage.
Best regards,
Michal
Commits
-------
21a5841 RedirectResponse tests
Discussion
----------
RedirectResponse tests
Hi,
This patch adds 100% test coverage for RedirectResponse class.
Best regards,
Michal
Commits
-------
04fd5f1 [Form] Fixed PropertyPath to not modify Collection instances (not even their clones)
Discussion
----------
[Form] Fixed PropertyPath to not modify Collection instances
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: #4670
Todo: -
---------------------------------------------------------------------------
by pocallaghan at 2012-08-31T14:20:52Z
As far as I can see the pull request does fix the issue, which makes sense based on the code change (I didn't know iterator_to_array existed, good call). One thing I would say, I'm not sure on the use in the change to the test case. It's not clear to me what additional protection this extra assertion gives, as both the old and new code seem to pass.
---------------------------------------------------------------------------
by bschussek at 2012-08-31T14:21:46Z
The new assertion is there because not even the old code (`clone`) was tested.
---------------------------------------------------------------------------
by stof at 2012-08-31T14:37:38Z
@bschussek but was it failing without the code change ?
---------------------------------------------------------------------------
by bschussek at 2012-08-31T22:12:00Z
@stof It was not, but I was unable to write a good test for the change within reasonable time. I added an explanatory comment instead.
They are part of the public API (as const are always public) but cannot be used at all from outside the class as the $pluralMap is private. The meaning of the indices is already documented in the array.
Commits
-------
7ef6295 Removed the unnecessary file include
Discussion
----------
Removed the unnecessary file include
As of Doctrine 2.2, the ORM annotations can be found by the autoloader directly.
This will also avoid breaking the testsuite in case the dev deps have not been installed as the ORM would not be available in this case (the tests relying on the ORM should already be skipped when it is not available). See #5402
Commits
-------
cb7e3f5 [Routing] added route compile check to identify a default value of a required variable that does not match the requirement
Discussion
----------
[Routing] added route compile check to identify a bad default value
BC break: yes but only for strange route definitions
See the exception message in code for the reasoning.
An exception is thrown for a __required__ variable that __has a default__ that __doesn't match__ the requirement.
So optional variables can of course still have a default that don't meet the requirement, which is useful.
This helps to identify useless route definitions at compile time instead of when generating or matching a URL.
Commits
-------
890aea2 FileLocatorInterface used in typehint instead of FileLocator
Discussion
----------
FileLocatorInterface used in typehint instead of FileLocator
---------------------------------------------------------------------------
by stof at 2012-08-30T22:09:39Z
@fabpot this makes sense (and it is BC)
---------------------------------------------------------------------------
by mvrhov at 2012-08-31T08:34:17Z
What's wrong with Interface hint? I always hint interface when available as this means that I can inject whatever class implementing that interface.
Commits
-------
3363832 extended phpdoc of ConfigurableRequirementsInterface
5f64503 [Routing] added test for disabled requirements check
4225869 [Routing] allow disabling the requirements check on URL generation
Discussion
----------
[Routing] allow disabling the requirements check on URL generation
This adds the possibility to disable the requirements check (`strict_requirements = null`) on URL generation.
To sum up, here the possibilities:
- `strict_requirements = true`: throw exception for mismatching requirements (most useful in development environment).
- `strict_requirements = false`: don't throw exception but return null as URL for mismatching requirements and log it (useful when you cannot control all params because they come from third party libs or something but don't want to have a 404 in production environment. it logs the mismatch so you can review it).
- `strict_requirements = null`: Return the URL with the given parameters without checking the requirements at all. When generating an URL you should either trust your params or you validated them beforehand because otherwise it would break your link anyway (just as with strict_requirements=false). So in production environment you should know that params allways pass the requirements. Thus you could disable the check on each URL generation for performance reasons. If you have 300 links on a page and each URL at least one param you safe 300 unneeded `preg_match` calls. I tested the performance in one of my projects. The rendering time of a single template that contains ~300 links with several params was reduced from avg. 46ms to avg. 42ms. That are 8.7% reduction in the twig layer where the links are created on each request. So this option combines the improved usability of strict_requirements=false with an additional increased performance.
---------------------------------------------------------------------------
by fabpot at 2012-08-30T13:40:38Z
Can you put the comment about all the possibilities you have mentioned here in the phpdoc for future reference? Thanks.
---------------------------------------------------------------------------
by Tobion at 2012-08-30T13:49:25Z
In `ConfigurableRequirementsInterface` or which phpdoc would you like to have it? Because `ConfigurableRequirementsInterface` already has it explained. But I can extend its description if thats what you mean.
---------------------------------------------------------------------------
by fabpot at 2012-08-30T13:50:40Z
The comment in the PR is more explicit and more detailed than the one in the interface. So, yes, basically, it would be great if you can move all the information in the interface phpdoc.
---------------------------------------------------------------------------
by Tobion at 2012-08-30T14:35:59Z
Done.