Commits
-------
731b28b [composer] add missing deps for FrameworkBundle
9c8f100 [composer] change ext/intl to the new ext-intl syntax
d535afe [composer] fix monolog-bridge composer.json, add more inter-component deps
9ade639 [composer] add composer.json
Discussion
----------
Composer
This PR adds a composer.json file for [composer](https://github.com/composer/composer) ([more info](packagist.org/about-composer)).
For discussion you can also go into #composer-dev on freenode and argue with naderman, seldaek and everzet.
---------------------------------------------------------------------------
by naderman at 2011/09/26 15:51:51 -0700
You haven't entered any keywords, they might come in handy when searching for packages on packagist.
But really this is just a +1 ;-)
---------------------------------------------------------------------------
by stof at 2011/09/26 16:12:21 -0700
See my comments on your previous (non-rebased) commit: f1c0242b5a
---------------------------------------------------------------------------
by igorw at 2011/09/27 00:04:36 -0700
Following dependencies do not have a composer.json yet: Twig, Doctrine (orm, dbal, common), swiftmailer.
Also missing from the standard edition: assetic, twig-extensions, jsm-metadata, SensioFrameworkExtraBundle, JMSSecurityExtraBundle, SensioDistributionBundle, SensioGeneratorBundle, AsseticBundle.
The point is, those can be added later on. Having the components composerized is already a leap forward. Also, doctrine depends on some symfony components, we've got to start somewhere.
---------------------------------------------------------------------------
by Seldaek at 2011/09/27 00:36:41 -0700
Also, just for information, the plan is to have `symfony/framework-bundle` be the "framework", with all dependencies to doctrine etc, though we should really only have strict requirements in there, and then in symfony-standard we ship a composer.json that requires the framework-bundle, doctrine-orm and things like that that are not essential to core. Otherwise people don't have a choice about what they use anymore.
Just a comment btw, the json is invalid, all / should be escaped. However json_decode is nice enough to parse those without complaining, browsers do too, even Crockford's json2.js does, so I'm not sure if we should privilege readability over strictness, since it seems nobody really cares about this escaping.
---------------------------------------------------------------------------
by igorw at 2011/09/27 00:41:39 -0700
So, I've implemented all of @stof's suggestions, except (for reasons stated above):
* doctrine to DoctrineBundle
* swiftmailer to SwiftmailerBundle
* twig to TwigBundle
* doctrine-common to Validator
* FrameworkBundle (what exactly does it depend on?)
---------------------------------------------------------------------------
by stof at 2011/09/27 00:52:31 -0700
@igorw at least HttpKernel, Routing, Templating, EventDispatcher, Doctrine Common (annotations cannot be disabled), Translator, Form (optional), Validator (optional), Console (optional). See the service definitions to see the others
@Seldaek FrameworkBundle does not depend on Doctrine, except for Common
---------------------------------------------------------------------------
by beberlei at 2011/09/27 03:15:34 -0700
What does the symfony/ or ext/ prefix control in composer?
---------------------------------------------------------------------------
by Seldaek at 2011/09/27 03:33:52 -0700
symfony/ is just the (mandatory) vendor namespace. Also ext/ has been renamed to ext- now, so it's not in any vendor, and should avoid potential issues.
---------------------------------------------------------------------------
by beberlei at 2011/09/27 05:07:03 -0700
@Seldaek Mandatory? So every package name is "vendor/package"? I like that because previously i thought package names are not namespaced, and thus clashes could occur between different communities easily.
---------------------------------------------------------------------------
by Seldaek at 2011/09/27 05:16:20 -0700
@beberlei: Mandatory. As of yesterday http://packagist.org/ will tell you you have an invalid package name if there's no slash in it. See 1306d1ca82 (diff-3)
Doctrine caches annotations. For methods, it uses PHP reflection and the getDeclaringClass() to create
a unique cache key. Unfortunately, if you have 2 classes that extend another one, the cache will be shared.
It's not a problem except that before this patch, the default route name was also cached (as the cache is serialized
after we changed the object). So, all other classes inherited this default route name. The fix is quite easy:
just don't change the read annotation object.
Commits
-------
03c7cfe UrlGenerator no longer appends '?' if query string is empty
Discussion
----------
UrlGenerator no longer appends '?' if query string is empty
If you generate a URL using null parameters (`array('foo' => null, 'bar' => null')`), `http_build_query` returns an empty string, resulting in a trailing `?` at the end of the generated URL.
This fixes that so that, if there are `$extra` params & `http_build_query` is empty, the URL is no longer appended.
---------------------------------------------------------------------------
by fabpot at 2011/07/22 10:15:26 -0700
Can you add unit tests?
---------------------------------------------------------------------------
by ericclemmons at 2011/07/22 10:52:21 -0700
Yes sir, will do.
-Eric Clemmons
Sent from my iPad Nano
On Jul 22, 2011, at 12:15 PM, fabpot<reply@reply.github.com> wrote:
> Can you add unit tests?
>
> --
> Reply to this email directly or view it on GitHub:
> https://github.com/symfony/symfony/pull/1773#issuecomment-1633515
---------------------------------------------------------------------------
by ericclemmons at 2011/07/22 11:55:30 -0700
**Added passing test.**
Currently `master` fails test:
```
1) Symfony\Tests\Component\Routing\Generator\UrlGeneratorTest::testUrlWithNullExtraParameters
Failed asserting that two strings are equal.
--- Expected
+++ Actual
@@ @@
-http://localhost/app.php/testing
+http://localhost/app.php/testing?
//tests/Symfony/Tests/Component/Routing/Generator/UrlGeneratorTest.php:114
```
Commits
-------
418d6a0 [Routing] Fix syntax error when dumping routes with single quotes in the requirements or pattern
2b5e22d [Routing] Fix ApacheDumper when a space appears in a default value
6c7f484 [Routing] Fix dumper so it doesn't print trailing whitespace
761724a [Routing] Adjust urlescaping rules, fixes#752
Discussion
----------
[Router] Bunch o' Fixes
The first commit changes the escaping rule to fix issues I had previously, and #752 as well, here's from the full commit message:
Only + and % are now encoded in generated routes, since they are the only characters that, if not encoded, could cause problems/conflicts when decoded. Namely + turns into a space, and % followed by numbers could do funky things.
The matcher decodes everything which works since nothing will have %NN without being escaped, and + are escaped as well.
Second commit is just a test fix for the first
Third and fourth are simply dumper escaping issues, nothing to argue about.
Note that all changes have had test cases added, and I spent a few hours torturing/testing all this stuff with both Apache and PHP dumpers, in many browsers, and with URLs as wonky as `/%01%02%03%04%05%06%07%08%0E%0F%10%11%12%13%14%15%16%17%18%19%1A%1B%1C%1D%1E%1F%20!%22$%25&%27%28%29*+,-./0123456789:;%3C=%3E@ABCDEFGHIJKLMNOPQRSTUVWXYZ%5B%5D%5E_%60abcdefghijklmnopqrstuvwxyz%7B|%7D~/baz` which essentially represent the 1-255 char range minus ? and #.
The only issues I really encountered after all the patches were applied is that Apache refuses to match `%22` (= `"`) and `*` in a url. I guess it's just because they're not allowed chars in windows paths, but | and < > works fine though. Anyway this works with the PHP dumper, and it didn't work either without my patches so it's not like I broke it, I'm just saying for the record.
Only + and % are now encoded in generated routes, since they are the only characters that, if not encoded, could cause problems/conflicts when decoded. Namely + turns into a space, and % followed by numbers could do funky things.
The matcher decodes everything which works since nothing will have %NN without being escaped, and + are escaped as well.
Commits
-------
49dd558 Couple more CS fixes
5a986bf Add $keysCount & minor CS fix
91f4097 [Routing] Better nesting for RouteCollections in dumped URL matcher classes With this change, a route prefixed with '/blogger' will be nested inside '/blog' (for example)
Discussion
----------
[Routing] Better nesting for RouteCollections in dumped URL matcher classes
Consider the following routing file:
_blog:
resource: "@AcmeDemoBundle/Resources/config/routing/BlogController.yml"
prefix: /blog
_blogger:
resource: "@AcmeDemoBundle/Resources/config/routing/BloggerController.yml"
prefix: /blogger
_bloggeroo:
resource: "@AcmeDemoBundle/Resources/config/routing/BloggerooController.yml"
prefix: /bloggeroo
_blogtest:
pattern: /blogtest
defaults: { _controller: AcmeDemoBundle:Blog:test }
login:
pattern: /login
defaults: { _controller: AcmeDemoBundle:Security:login }
Note: Each imported file contains two simple blog/blogger/bloggeroo routes (e.g. blog_index & blog_view)
With this change, the cached URL matcher looks something like this:
```php
<?php
class appprodUrlMatcher
{
public function match($pathinfo)
{
$allow = array();
if (0 === strpos($pathinfo, '/blog')) {
// blog_index
if (preg_match('#^/blog/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
// blog_view
if (preg_match('#^/blog/(?P<year>\d{4})/(?P<month>\d{2})/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
if (0 === strpos($pathinfo, '/blogger')) {
// blogger_index
if (preg_match('#^/blogger/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
// blogger_view
if (preg_match('#^/blogger/(?P<year>\d{4})/(?P<month>\d{2})/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
if (0 === strpos($pathinfo, '/bloggeroo')) {
// bloggeroo_index
if (preg_match('#^/bloggeroo/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
// bloggeroo_view
if (preg_match('#^/bloggeroo/(?P<year>\d{4})/(?P<month>\d{2})/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {...}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
// _blogtest
if ($pathinfo === '/blogtest') {...}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
// login
if ($pathinfo === '/login') {...}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
}
```
As far as I can see, you can throw the 404 earlier (as I've done), as by nesting all these possibilities, there is no chance another route will be matched outside the parent if statement.
---------------------------------------------------------------------------
by stloyd at 2011/06/11 08:37:15 -0700
You should follow Symfony [CS rules] (http://symfony.com/doc/current/contributing/code/standards.html), also you should modify tests.
---------------------------------------------------------------------------
by lmcd at 2011/06/11 08:46:32 -0700
Besides the inline `continue` statement, I can't see spot any other CS violations. I'm on to the updated tests.
---------------------------------------------------------------------------
by fabpot at 2011/06/13 02:19:14 -0700
What if you have this:
_blog:
resource: "@AcmeDemoBundle/Resources/config/routing/BlogController.yml"
prefix: /blog
login:
pattern: /login
defaults: { _controller: AcmeDemoBundle:Security:login }
_blogger:
resource: "@AcmeDemoBundle/Resources/config/routing/BloggerController.yml"
prefix: /blogger
You cannot send 404 early because the same `blog` prefix is used in two different places. I can see many things that will break with this patch. I'm pretty sure we can enhance the performance of the existing code but let's do that after 2.0.
---------------------------------------------------------------------------
by lmcd at 2011/06/13 05:04:03 -0700
@fabpot: The code was designed so it would work in these kind of circumstances. I ran the dumper against the routing file you provided and this is the output below.
I see no reason why you can't throw the 404 earlier, as by nesting everything with similar prefixes, we're eliminating the possibility that a route will ever be matched outside of that if statement.
```php
public function match($pathinfo)
{
$allow = array();
if (0 === strpos($pathinfo, '/blog')) {
// blog_index
if (preg_match('#^/blog/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
...
}
// blog_view
if (preg_match('#^/blog/(?P<year>\d{4})/(?P<month>\d{2})/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
...
}
if (0 === strpos($pathinfo, '/blogger')) {
// blog_index
if (preg_match('#^/blogger/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
...
}
// blog_view
if (preg_match('#^/blogger/(?P<year>\d{4})/(?P<month>\d{2})/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
...
}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
// login
if ($pathinfo === '/login') {
...
}
throw 0 < count($allow) ? new MethodNotAllowedException(array_unique($allow)) : new ResourceNotFoundException();
}
```
---------------------------------------------------------------------------
by fabpot at 2011/06/13 07:48:49 -0700
Can you add some more tests to ensure that everything works fine and that we won't have regressions later on? Thanks.
* lmcd/master:
$code referenced but not defined in compileRoute()
[Routing] Optimised the PHP URL matcher dumper The cached URL matcher classes contain some unneeded logic. Consider the following example:
The cached URL matcher classes contain some unneeded logic. Consider the following example:
if (0 === strpos($pathinfo, '/Blog')) {
// blog_index
if (0 === strpos($pathinfo, '/Blog') && preg_match('#^/Blog/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
return array_merge($this->mergeDefaults($matches, array ( '_action' => 'index',)), array('_route' => 'blog_index'));
}
}
The 2nd strpos is not required, as we have already satisfied this condition in the parent if statement.
My change will produce the following code for the same routing setup::
if (0 === strpos($pathinfo, '/Blog')) {
// blog_index
if (preg_match('#^/Blog/(?P<slug>[^/]+?)$#x', $pathinfo, $matches)) {
return array_merge($this->mergeDefaults($matches, array ( '_action' => 'index',)), array('_route' => 'blog_index'));
}
}
Here are the new simplified rules:
* Required cache warmers are *always* executed when the Kernel boots for the first time;
* Optional cache warmers are *only* executed from the CLI via cache:warmup
These new rules means that all the configuration settings for the cache
warmers have been removed. So, if you want the best performance, remember to
warmup the cache when going to production.
This also fixed quite a few bugs.
* Chekote/xmlfileloader_overload_refactor:
Refactored the processing of each individual node into it's own method, enabling easier overloading of behavior for Bundles such as FriendsOfSymfony/RestBundle
* alexandresalome/feat-routing-exceptions:
[Routing] Fix the exception inheritance + Add the LICENCE block in new files
[Routing] Change the Exception namespacing + base class for every exception + update PHPDoc
[Routing] Add specific exceptions for the UrlGenerator
When generating URL, thrown exceptions are InvalidArgumentException and
distinction of errors is quite difficult. This modification brings different
exceptions for different cases
The _scheme requirement can be used to force routes to always match one given scheme
and to always be generated with the given scheme.
So, if _scheme is set to https, URL generation will force an absolute URL if the
current scheme is http. And if you request the URL with http, you will be redirected
to the https URL.
This improves, for example, the exception one would receive if they tried to import a resource from a bundle that doesn't exist.
Previously, the deep "bundle is not activated" exception would be thrown. That has value, however there is no indication of where
the exception is actually occurring.
In this new implementation, we throw an exception that explains exactly which resource, and from which source resource, cannot be
loaded. The deeper exception is still thrown as a nested exception.
Two caveats:
* The `HttpKernel::varToString` method was replicated
* This introduces a new `Exception` class, which allows us to prevent lot's of exceptions from nesting into each other in the case
that some deeply imported resource cannot be imported (each upstream import that fails doesn't add its own exception).
* subsven/master:
re-add filename based directory filter to the AnnotationDirectoryLoader, now restricting to *.php files and therefore disregarding e.g. SVN metadata files
revert adding filename based filter to the directory resource
Eleminate the need to manually clear the cache if a new controller file containing routing annotations is added - part II * add unit tests * introduce filename filter to DirectoryResource (to restrict change monitoring to a subset of files) * modify AnnotationDirectoryLoader.php to use filename filter set on Controller\.php$
* add unit tests
* introduce filename filter to DirectoryResource (to restrict change monitoring to a subset of files)
* modify AnnotationDirectoryLoader.php to use filename filter set on Controller\.php$