Commits
-------
2a90871 [Console] Removed previously introduced BC break.
90a2a6e [Console] Undecorated formatter must update style stack too.
bd7e01a [Console] Fixed output formatter test broken by new implementation.
a1add4b [Console] Updated output formatter to use style stack.
4f298dd [Console] Added formatter style stack.
93ffe54 [Console] Added getters to output formatter style (and its interface).
48e6b49 [Console] Updated formatter test to match styles bug fix.
ad334b6 [Console] Fixed empty style appliance.
31d5fe5 [Console] Fixed output formatter docblock.
Discussion
----------
[Console] Fixes formatter nested style appliance.
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
When outputing styled text in the console, you sometimes face to a confusing behavior: style tags cannot be nested. If tou try something like `<fg=blue>Hello <fg=red>world</fg=red>!</fg=blue>`, the trailing `!` will not be styled.
This PR introduce a new FormatterOutputStyleStack to keep open/closed styles informations up-to-date. It slightly changes OutputFormatter implementation which no longer uses `OutputFormatterStyle::apply()` method, but the new `OutputFormatterStyle::getTerminalSequence()`.
**Question:** I don't une `OutputFormatterStyleInterface` but `OutputFormatterStyle` to type `OutputFormatterStyleStack` methods arguments (to avoid BC break on the interface). Do you think it's right?
**Notice:** I also needed to fix some tests broken by new implementation.
---------------------------------------------------------------------------
by stof at 2012-03-16T10:27:56Z
Adding new methods in an interface is a BC break for people implementing it
---------------------------------------------------------------------------
by jfsimon at 2012-03-16T10:33:21Z
@stof indeed... this is a problem, should I remove them? If I do so, I should use `OutputFormatterStyle` instead of the interface to type arguments in `OutputFormatterStyleStack` right?
Commits
-------
8a0e6d2 [HttpFoundation] Update changelog.
4fc04fa [HttpFoundation] Renamed MetaBag to MetadataBag
2f03b31 [HttpFoundation] Added the ability to change the session cookie lifetime on migrate().
39141e8 [HttpFoundation] Add ability to force the lifetime (allows update of session cookie expiry-time)
ec3f88f [HttpFoundation] Add methods to interface
402254c [HttpFoundation] Changed meta-data responsibility to SessionStorageInterface
d9fd14f [HttpFoundation] Refactored for moved tests location.
29bd787 [HttpFoundation] Added some basic meta-data to Session
Discussion
----------
[2.1][HttpFoundation] Added some basic meta-data to Session
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
References the following tickets: #2171
Todo: -
Session data is stored as an encoded string against a single id. If we want to store meta-data about the session, that data has to be stored as part of the session data to ensure the meta-data can persist using any session save handler.
This patch makes it much easier to determine the logic of session expiration. In general a session expiry can be dealt with by the gc handlers, however, in some applications more specific expiry rules might be required.
Session expiry may also be more complex than a simple, session was idle for x seconds. For example, in Zikula there are three security settings, Low, Medium and High. The rules for session expiry are more complex as under the Medium setting, a session will expire after x minutes idle time, unless the rememberme option was ticked on login. If so, the session will not idle. This gives the user some control over their experience. Under the high security setting, then there is no option, sessions will expire after the idle time is reached and login the UI has the rememberme checkbox removed.
The other advantage is that under this methodology, there can be a UI experience on expiry, like "Sorry, your session expired due to being idle for 10 minutes".
Keeping in the spirit of Symfony2 Components, I am seeking to make session handling flexible enough to accommodate these general requirements without specifically covering expiration rules. It would mean that it would be up to the implementing application to specifcally check and expire session after starting it.
Expiration might look something like this:
$session->start();
if (time() - $session->getMetadataBag()->getLastUpdate() > $maxIdleTime) {
$session->invalidate();
throw new SessionExpired();
}
This commit also brings the ability to change the `cookie_lifetime` when migrating a session. This means one could move from a default of browser only session cookie to long-lived cookie when changing from a anonymous to a logged in user for example.
$session->migrate($destroy, $lifetime);
---------------------------------------------------------------------------
by drak at 2012-03-30T18:18:43Z
@fabpot I have removed [WIP] status.
---------------------------------------------------------------------------
by drak at 2012-03-31T13:34:57Z
NB: This PR has been rebased and the tests relocated as per recent master changes.
---------------------------------------------------------------------------
by drak at 2012-04-03T02:16:43Z
@fabpot - ping
Commits
-------
56c1e31 performance improvement in PhpMatcherDumper
Discussion
----------
performance improvement in PhpMatcherDumper
Tests pass: yes
The code generation uses a string internally instead of an array. The array wasn't used for random access anyway.
I also removed 4 unneeded iterations this way (when imploding, when merging and twice when applying the extra indention). A `preg_replace` could also be saved under certain circumstances by moving it.
And there was a small code errror in line 139.
Commits
-------
f1f1494 Added an exception when passing an invalid object to ApcClassLoader
f5cb167 [ClassLoader] Added a DebugClassLoader using composition
0e54a22 Updated the changelog
eae772e [ClassLoader] Added an ApcClassLoader
4d1333f Changed the test autoloading to use the new autoloader
09850bd [ClassLoader] Added a simplified PSR-0 ClassLoader
Discussion
----------
Autoloader refactoring
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: [](http://travis-ci.org/stof/symfony)
As discussed in #3623, I added a new ClassLoader instead of modifying the UniversalClassLoader, to be able to use the method names without BC concerns. The new class works the same than the composer autoloader regarding the handling of fallbacks, to be able to reuse namespace maps generated by composer.
```php
<?php
// autoload.php
require_once __DIR__.'/vendor/symfony/class-loader/Symfony/Component/ClassLoader/ClassLoader.php';
$loader = new Symfony\Component\ClassLoader\ClassLoader();
$map = require __DIR__.'/vendor/.composer/autoload_namespaces.php';
$loader->addPrefixes($map);
$loader->register();
```
Differences with the composer class loader:
- Composer's ``add`` method is named ``addPrefix`` in the Symfony ClassLoader
- the methods related to the class map are removed as Symfony has a separate laoder for class maps
- the ``addPrefixes`` method is added, accepting a namespace map.
I also added a new ApcClassLoader which uses composition instead of inheriting from a class loader, which makes it far more easier to reuse (we could wrap a Composer autoloader with it for instance).
```php
<?php
$composerLoader = require __DIR__.'/vendor/.composer/autoload.php';
// no need to require the file manually as Composer already registered its autoloader
$cachedLoader = new Symfony\Component\ClassLoader\ApcClassLoader('autoload.my_app', $composerLoader);
$cachedLoader->register();
// unregister the Composer autoloader as we wrapped it in the ApcClassLoader
$composerLoader->unregister();
```
TODO:
- refactor the Debug class loader to use composition too to be able to support different class loaders
---------------------------------------------------------------------------
by fabpot at 2012-04-02T16:31:28Z
Can you update the CHANGELOG and the UPGRADE file accordingly?
---------------------------------------------------------------------------
by stof at 2012-04-02T16:47:43Z
I added a note in the CHANGELOG. There is nothing to add in the UPGRADE file as the change is fully BC (I did not change the UniversalClassLoader at all so it can still be used).
I'm working on the Debug loader right now so please wait a bit before merging
---------------------------------------------------------------------------
by stof at 2012-04-02T17:12:11Z
Here is a new DebugClassLoader using composition too. this way, it is able to support the UniversalClassLoader, the ApcUniversalClassLoader (without dropping the use of APC as done previously), the new ClassLoader, the new ApcClassLoader and even the composer autoloader.
I'm not sure about the use of ``method_exists`` as it could break if an autoloader implements a protected ``findFile`` method (crappy PHP 😢) but hardcoding the supported classes would be a pain and requiring an interface would make the autoloaders more difficult to use (as the interface would need to be required first) and would drop the support of the composer autoloader.
Commits
-------
dbab7e1 [TwigBridge] Added a TwigEngine in the bridge
Discussion
----------
[TwigBridge] Added a TwigEngine in the bridge
This TwigEngine implements the interface available in the component.
the TwigBridge in TwigBundle now extends this class and provides only
the additional methods for the FrameworkBundle interface.
This will allow people to support the PhpEngine and Twig in their code more easily when using the component as they don't need to reimplement the class.
I originally thought about when I helped @dragoonis to integrate the Templating component in the PPI framework 2.0 as he talked about adding the support for Twig later too.
Commits
-------
93848be moved event dispatcher classes to the EventDispatcher component
Discussion
----------
moved event dispatcher classes to the EventDispatcher component
I have moved the two specialized event dispatcher classes from the FrameworkBundle bundle to the EventDispatcher component.
It makes them reusable outside the Symfony full-stack framework as they can be quite useful (people like @drak has a use for them for instance).
It makes the event dispatcher component *optionally* rely on DependencyInjection and HttpKernel for these classes, like what we have in other components.
What do you think?
---------------------------------------------------------------------------
by stof at 2012-04-02T16:02:01Z
you forgot to update the ``suggests`` section of the composer.json file. Otherwise, it is fine
---------------------------------------------------------------------------
by drak at 2012-04-02T16:06:34Z
w00t, thanks @fabpot!
I'm biased of course, but the optional dependency to `DependencyInjection` is a much more logical dependency that the previous one, to `FrameworkBundle`. It immediately shows of the DIC also.
---------------------------------------------------------------------------
by schmittjoh at 2012-04-02T16:06:39Z
+1, the "no dependencies" should really be changed to "minimal dependencies"
---------------------------------------------------------------------------
by henrikbjorn at 2012-04-02T16:09:48Z
@schmittjoh should properly be "minimal optional dependencies" otherwise people will be scared i think :) And reading "minimal dependencies" reminds me when trying to use a zf1 component standalone.
+1
---------------------------------------------------------------------------
by drak at 2012-04-02T16:12:57Z
@schmittjoh Honestly speaking, some of the Sf2 components do already have a few dependencies with each other, but where there are, they all minimal and rather logical anyway. It's not like some libraries I could mention where one lib required about 10 deps which each rely on a bunch of other things - in the end, you need everything just for one feature :-P
---------------------------------------------------------------------------
by stof at 2012-04-02T16:14:28Z
@drak there was not optional dependency to FrameworkBundle previously as no code of the component was using FrameworkBundle (the dependency was the other way)
---------------------------------------------------------------------------
by drak at 2012-04-02T16:18:31Z
@stof, yeah but I was talking about other components like Security, DependencyInjection, HttpKernel for example which has a mix of optional and required deps. It's not true to say all Sf2 components are 100% standalone in all cases.
This TwigEngine implements the interface available in the component.
the TwigBridge in TwigBundle now extends this class and provides only
the additional methods for the FrameworkBundle interface.
Unlike the ApcUniversalClassLoader, ApcClassLoader uses composition,
meaning it can be used to wrap any object providing a findFile($class)
method. Both the UniversalClassLoader and the new ClassLoader follow
this convention. It can also be used to wrap the Composer autoloader.
The new ClassLoader does not differentiate namespaced classes and
PEAR-like classes like the UniversalClassLoader does as the PEAR
format is a subset of PSR-0.
The new loader registers fallbacks by adding a location for an empty
prefix, as done in Composer. This allows using namespaces map generated
by Composer without any special processing on them.
Commits
-------
dcf82d7 [Finder] Added sortBy options based on accessed, changed and modified times
Discussion
----------
[Finder] Added sortBy options based on accessed, changed and modified times
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
Commits
-------
925b65d updated reference to tests
Discussion
----------
updated reference to tests
this is basically the format used in all other components.
however i am not sure if it really makes sense to list ``phpunit -c src/Symfony/Component/HttpFoundation/``, since this relative path could be confusing for anyone using the standalone components. But even if using ``symfony/symfony`` the path is wrong relative to the location of the README.md.
Commits
-------
24a0d0a [DependencyInjection] Support Yaml calls without arguments
Discussion
----------
[DependencyInjection] Support Yaml calls without arguments
This is a very important option which allows the cookie lifetime to be changed on migrate.
For example when a user converts from an anonymous session to a logged in session one might
wish to change from a persistent cookie to browser session (e.g. a banking application).
This commit allows applications to know certain meta-data about the session
Session storage is designed to only store some data against a session ID
so this method is necessary to be compatible with any session handler, including
native handlers.
Commits
-------
8dd2c27 [HttpFoundation] Further micro-optimization.
54c5d5e [HttpFoundation] Micro-optimisation.
Discussion
----------
[HttpFoundation] Micro-optimisation.
Ref #3729
---------------------------------------------------------------------------
by robocoder at 2012-03-30T11:45:02Z
If you pre-flip your $validOptions arrays, you can use isset() instead of in_array() in the loop.
This changes the performance from O(m * n) to O(m).
---------------------------------------------------------------------------
by drak at 2012-03-30T11:53:24Z
@robocoder What is the expense of the array_flip though?
---------------------------------------------------------------------------
by robocoder at 2012-03-30T11:56:21Z
Why would you use array_flip if the array doesn't change? Change $validOptions = array('x', 'y', ...) to $validOptions = array('x' => 0, 'y' => 0, ...), then change the in_array() to use isset().
---------------------------------------------------------------------------
by stof at 2012-03-30T11:57:08Z
@drak a loop. But it will be done only once before the other loop so it will be O(n + m) instead of O(m * n)
---------------------------------------------------------------------------
by drak at 2012-03-30T12:00:47Z
Ok :)
Commits
-------
234ce4d [PhpUnit] Fix the path to the boostrap files in the components
Discussion
----------
[PhpUnit] Fix the path to the boostrap files in the components
