Commits
-------
6f9eda9 [Form][Validator] Fixed generation of HTML5 pattern attribute based on Assert\Regex to remove delimiters.
Discussion
----------
[Form][Validator] Fixed generation of HTML5 pattern attribute based on Assert\Regex by removing delimiters or using a new option: htmlPattern.
Hopefully, this time is the good one…
* Fixes: [#3766, #4077, #4513, #4520, #4521]
* Bug fix: yes
* Feature addition: yes
* BC break: no
* Symfony2 tests pass: yes
In Issue #3766, it was asked that Assert\Regex generates HTML5 pattern attribute.
It was done in PR #4077, but the generated Regex is in delimited format which is not supported by HTML5.
Hence, `/[a-z]+/` would be converted to `[a-z]+`.
If flags are specified like in `/[a-z]+/i`, it cannot be converted and pattern validation will be disabled client-side. If is however now possible, using a new option, `htmlPattern`, to specify the pattern you want to be used.
Example:
```php
<?php
/**
* @Assert\Regex(pattern="/^[0-9]+[a-z]*$/i", htmlPattern="^[0-9]+[a-zA-Z]*$")
*/
private $civic_number;
```
**Note**: [Documentation](http://symfony.com/doc/current/reference/constraints/Regex.html) should be updated accordingly.
---------------------------------------------------------------------------
by lavoiesl at 2012-06-08T15:45:17Z
God, I just found out you can "add more commits to this pull request by pushing to the master branch on lavoiesl/symfony"…
---------------------------------------------------------------------------
by travisbot at 2012-06-08T15:50:31Z
This pull request [passes](http://travis-ci.org/symfony/symfony/builds/1568634) (merged 2d767b41 into b84b46ba).
---------------------------------------------------------------------------
by petajaros at 2012-07-04T14:23:16Z
Anything new about this issue?
---------------------------------------------------------------------------
by lavoiesl at 2012-07-04T16:25:43Z
Alright, tests are passing using `phpunit -c phpunit.xml.dist --filter 'RegexValidatorTest'`. @travisbot reports errors because he can’t even start the tests due to dependencies, which is not related
---------------------------------------------------------------------------
by vicb at 2012-07-04T16:31:13Z
It should be ready to merge when you have taken the last comments into account. thanks.
---------------------------------------------------------------------------
by lavoiesl at 2012-07-04T16:39:05Z
So it seems this PR will finally pass, thanks a lot.
---------------------------------------------------------------------------
by vicb at 2012-07-04T17:03:35Z
Thank you for this PR and the changes.
---------------------------------------------------------------------------
by fabpot at 2012-07-04T17:10:20Z
@lavoiesl Can you squash your commits before I merge? Thanks.
---------------------------------------------------------------------------
by lavoiesl at 2012-07-04T17:25:18Z
There. I also left trace of some commits I did.
Thanks
[Validator] Added delimiter escaping to Validator\Constraints\Regex::getNonDelimitedPattern
[Form][Validator] Added htmlPattern option for Regex Validation.
[Validator] Fixed Validator\Constraints\Regex::getNonDelimitedPattern variable declarations
[Validator] Fixed tests for Regex htmlPattern option (instead of html_pattern)
[Validation] tweaked generation of pattern to include .* when not anchors are present. Also removed the exception and made getNonDelimitedPattern private
Commits
-------
0555913 [FrameworkBundle] Allow using kernel parameters in routes
Discussion
----------
[FrameworkBundle] Allow using kernel parameters in routes
Kernel parameters can now be used at any position in patterns, defaults and requirements.
Relates to: #3316, #3276
**Differences from 3316:**
- The substitution is now done in the `Router`,
- 3316 uses `$container->getParameterBag()` which is not part of the `ContainerInterface`. The way it been solved in this PR is that some code have been duplicated inside the `Router`, see `resolveString()`.
**BC break:**
Before this PR, nonexistent parameters would have be silently ignored (ie `%idontexist%` would not have been replaced). After this PR, they will throw an exception. However you can escape the value (ie `%%idontexist%%` will be accepted and replaced by `%idontexist%`).
_This behavior is not mandatory and can be reverted if needed. However this keeps the router more consistent with the DI_.
Any feedback ? @helmer @Koc
---------------------------------------------------------------------------
by Seldaek at 2012-07-04T12:40:08Z
👍 for consistency.
---------------------------------------------------------------------------
by helmer at 2012-07-04T13:07:11Z
+1 a much better solution to the problem than mine, closing #3316
---------------------------------------------------------------------------
by Tobion at 2012-07-04T13:21:59Z
How about escaping kernel params with `\%idontexist%` as I suggested it for route params in 4563?
---------------------------------------------------------------------------
by stof at 2012-07-04T13:28:55Z
@Tobion this would not be consistent with the way DI parameters are escaped elsewhere
---------------------------------------------------------------------------
by Koc at 2012-07-04T14:24:25Z
Looks good for me, thanx.
Commits
-------
30ca007 Changed the minimum stability to dev for the testsuite
Discussion
----------
Changed the minimum stability to dev for the testsuite
This will fix the testsuite as the default stability changed to stable in composer today
Commits
-------
18f8fc7 [Finder] tests for Yaml class
Discussion
----------
[Finder] tests for Yaml class
Bug fix: no
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: [](http://travis-ci.org/gajdaw/symfony)
Fixes the following tickets: -
Todo: -
License of the code: MIT
Documentation PR: -
Additional tests improving tests coverage.
Commits
-------
3f9e8ff [ClassLoader] made ClassCollectionLoader::load() automatically include class dependencies
6f4d281 [ClassLoader] added missing support for PHP 5.4 traits
Discussion
----------
Classloader optimization
The first commit fixes support for PHP 5.4 trait.
The second one does several things:
* it optimizes the recent merge so that the reflection class instance is only loaded once;
* we use the fact that we now get all class dependencies to automatically add all class dependencies to the map.
---------------------------------------------------------------------------
by fabpot at 2012-07-03T17:26:46Z
I've updated to take into accounts traits.
---------------------------------------------------------------------------
by bamarni at 2012-07-04T11:58:57Z
great job 👍
I can't see it in the diff as this part hasn't changed, but somewhere in the autoReload block there is :
```
if ($meta[1] != $classes) {
$reload = true;
}
```
It should be array_unique($classes), otherwise the file would be perpetually regenerated in autoReload mode when the input contains duplicate, because they're implicitely removed when dumping the files.
---------------------------------------------------------------------------
by fabpot at 2012-07-04T13:20:04Z
@bamarni I've added an `array_unique` call at the top (this bug existed before by the way).
Commits
-------
51b610f [Profiler] fix typehint
Discussion
----------
[Profiler] fix typehint
---------------------------------------------------------------------------
by fabpot at 2012-07-03T10:23:25Z
The profiler only works with Twig anyway.
---------------------------------------------------------------------------
by Tobion at 2012-07-03T10:29:18Z
Right. But why does he have this error: f47b9a6625 (commitcomment-1532164)
And since the class only uses methods of the general interface, I thought this makes it more reliable.
---------------------------------------------------------------------------
by fabpot at 2012-07-03T10:37:26Z
It was unrelated and I fixed that problem already.
Commits
-------
c40a4e5 [HttpFoundation] fix query string normalization
f9ec2ea refactored test method
0880174 [HttpFoundation] added failing tests for query string normalization
Discussion
----------
[HttpFoundation] fix query string normalization
This fixes the query string normalization. There were several problems in it (see test cases that I added).
The main issue, that first catched my eye, was that the query string was urldecoded before it was exploded by `=`. See old code: `explode('=', rawurldecode($segment), 2);`. This means an encoded `=` (`%3D`) would falsely be considered a separator and thus lead to complete different parameters. The fixed test case is at `pa%3Dram=foo%26bar%3Dbaz&test=test`.
---------------------------------------------------------------------------
by Tobion at 2012-07-04T02:21:25Z
cc @simensen considering your PR 4711
Commits
-------
d37003e [HttpFoundation] small fixes in Request
Discussion
----------
[HttpFoundation] small fixes in Request
phpdoc fixes,
making http_build_query explicit
fixing query string of '0', that was ignored.
Unfortunately this '0' problematic is omnipresent because PHP makes it so easy to get wrong (as it is converted to boolean false). I don't know how often I fixed such issue already.
Commits
-------
45219ef [Process] Add default xampp path to the list of possible paths to check
28e1313 [Process] Clean-up/simplify code
56f473a [Process] Add extra dirs argument to the executable finder to allow searching more dirs
Discussion
----------
ExecutableFinder changes
It cleans up stuff a bit and adds a guess for xampp users that wouldn't have php in their path
Commits
-------
eda439f [EventDataCollector] Display a better message when no events have been recorded
6b87981 [TimeDataCollector] Do not throw an exception when no events are recorded
Discussion
----------
[Profiler] Better support for collector in a production env
relates to #3706.
With this PR it is possible to:
- enable only the profiler in a production environment - the wdt being disabled you have to switch to the development environment to inspect the collected data,
- enable both the profiler and the wdt in a production environment (the use case form #3706).
@jmikola would this solve your use case ?
Commits
-------
45d967e [Form] Fixed: Empty forms can be compound too
Discussion
----------
[Form] Fixed: Empty forms can be compound too
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: #4480
Todo: -
Commits
-------
26a1e0b [ClassLoader] ordered ClassCollectionLoader writing to avoid redeclaration at runtime
Discussion
----------
[HttpKernel] allowed classes to compile to be prepended
I had an issue when registering JMSDIExtraBundle before the frameworkBundle, because the bundle is adding a class to compile wich extends a class to compile added by the frameworkbundle (https://github.com/schmittjoh/JMSDiExtraBundle/blob/master/HttpKernel/ControllerResolver.php#L39).
In my kernel, the bundle was registered before the frameworkbundle, if it's the case, the class is writed before the symfony core class in the cache file, so it will trigger the autoloader to load the symfony core class, then we'll have a fatal error because we're declaring 2 times the same class.
I'm suggesting to add a way to prepend the classes to compile added by an extension, this way we could force classes from the core bundle to pe prepended, and avoid this kind of error with other bundles adding classes extending core classes or implementing core interface. I've also added it to the frameworkbundle.
---------------------------------------------------------------------------
by travisbot at 2012-07-01T12:32:28Z
This pull request [fails](http://travis-ci.org/symfony/symfony/builds/1750090) (merged a989d35c into a1b73887).
---------------------------------------------------------------------------
by schmittjoh at 2012-07-01T12:45:46Z
This doesn't sound really failure proof, better would be to resolve the compiled classes in a way that handles the dependencies gracefully regardless of the order that they are registered in.
---------------------------------------------------------------------------
by bamarni at 2012-07-01T13:03:15Z
yes that would be much cleaner, even if it's not perfect, I don't have any other examples, but I think in most of situations this error happens because of the frameworkbundle late registration?
---------------------------------------------------------------------------
by bamarni at 2012-07-01T23:07:24Z
I've added an automatic reordering as suggested. Now I don't get the error whatever the order of FrameworkBundle and JMSDIExtraBundle is.
---------------------------------------------------------------------------
by vicb at 2012-07-02T10:02:33Z
Does the algo really works ?
Let's say I have the following hierarchy `A < B < C` and the classes added in the following order; `C, B, A`. I believe that the algo would return `B, A, C` instead of `A, B, C`.
An other question: should the dependency resolution be implemented in the `ClassCollectionLoader` rather than in the compiler pass ?
---------------------------------------------------------------------------
by vicb at 2012-07-02T10:23:11Z
@bamarni could you confirm the issue with the algo ? If it is confirmed, do you have time to work on a fix ?
---------------------------------------------------------------------------
by bamarni at 2012-07-02T11:05:53Z
the algo looks correct, there is a loop checking dependencies and appending the parent classes at the begining so it should also work with A > B > C, but I'll check and add a unit test.
You're right about the location it could belong there if we want a general fix, should I put it in the ClassLoader component?
---------------------------------------------------------------------------
by vicb at 2012-07-02T11:28:14Z
Yep please move the code to the ClassLoader.
You could also add some cache mechanism:
- ReflectionClass could be cached (they might be use mulitple times when some dependencies exist and are also used by the class loader)
- May be you could also cache the hierarchy of the classes
---------------------------------------------------------------------------
by bamarni at 2012-07-02T17:25:08Z
@vicb : indeed there were something wrong with the algo when it needed to handle more than 1 level of dependency, it's fixed, I've added a few test case and it passes.
Caching the dependencies found with reflection would be useful if people are dumping several files with some similar classes in the same process, but I don't see why we would need to cache the hierarchy?
---------------------------------------------------------------------------
by vicb at 2012-07-02T18:29:06Z
@bamarni I was referring to a local cache (ie local variable).
- On the first iteration, the code instantiate a reflection class,
- On other iteration you would instantiate again a reflection class for a subset of the classes
- ...
- In the end the former code instantiate again a reflection class for each class in order to get the file name, ...
I was also thinking of an other algo: count the number of parent classes for each classes to include and order classes according to this number. This would be a single loop only, what do you think ?
---------------------------------------------------------------------------
by bamarni at 2012-07-02T19:32:34Z
hum good idea counting the parents is cleaner, even though it looks enough, would it also make sense to treat interfaces and classes separately? I'm thinking about a file with all the interfaces at the begining (ordered by number of parents desc), then the classes would be appended (ordered by number of parents desc too).
---------------------------------------------------------------------------
by vicb at 2012-07-03T07:05:07Z
There is no real benefit in making the interfaces appear first but you can do it if it doesn't make the code more complex.
Please also add the `@throws` annotation to the methods.
---------------------------------------------------------------------------
by bamarni at 2012-07-03T09:37:28Z
@vicb : I've changed it to use the parents count, looks good to me.
---------------------------------------------------------------------------
by bamarni at 2012-07-03T12:23:46Z
changed
---------------------------------------------------------------------------
by bamarni at 2012-07-03T12:42:29Z
fixed
Sylvain Lorinet
Fabien Potencier
Sébastien Lavoie
Włodzimierz Gajda
Christophe Coevoet
Anthon Pang
Victor Berchet
Douglas Greenshields
Tobias Schultze
Bernhard Schussek
Jordi Boggiano
Jeremy Mikola
Frédéric Hardy
