Commits
-------
3223c5a Removed now useless test
21cf0ac Backported new behaviour from PR #2148 and removed check for interface at run-time
8b240d4 Implementation of kernel.event_subscriber tag for services.
Discussion
----------
Added missing kernel.event_subscriber tag (closes#2012)
This PR adds a ``kernel.event_subscriber`` tag which allows to register services as event subscribers in the same way ``kernel.event_listener`` allows to register them as event listeners.
The service is still lazy loaded and the DIC does not need to be recompiled for every modification in the service's code.
There is one important thing to remember:
If the service is created by a factory, the class parameter **MUST** reflect the real class of the service, although it is not needed at the moment for the DIC. For that issue, we could either forbid services created by factories or add a note to the documentation.
This PR closes#2012.
---------------------------------------------------------------------------
by jalliot at 2011/08/24 06:42:18 -0700
I'm not sure the test is good enough so feel free to add some more.
---------------------------------------------------------------------------
by jalliot at 2011/08/25 03:46:20 -0700
I re-implemented the check for EventSubscriberInterface in ContainerAwareEventDispatcher because I think the overhead is minimum and it allows to use this method even without the tag (at run-time).
I also added some tests for RegisterKernelListenersPass.
---------------------------------------------------------------------------
by stof at 2011/09/04 02:42:00 -0700
@jalliot Your branch conflicts with the current master. could you rebase it ?
---------------------------------------------------------------------------
by jalliot at 2011/09/04 02:57:03 -0700
Rebased
---------------------------------------------------------------------------
by jalliot at 2011/09/13 02:19:46 -0700
@fabpot What do you think about this PR? At the moment, the subscribers are not really usable in Symfony2 because of the lack of this tag.
---------------------------------------------------------------------------
by fabpot at 2011/09/13 04:17:46 -0700
I don't like subscribers. There are other PRs on adding more support for them, but the reality is that they are complex for no added benefit. I'm wondering if it wouldn't be better to just remove them altogether.
---------------------------------------------------------------------------
by jalliot at 2011/09/13 04:38:20 -0700
@fabpot Well I prefer listeners too but I think that if Symfony2 does support subscribers (which it does at the moment), it should do it properly and completely, thus allowing to register subscriber services like here or to register several methods for one same event like in #2148.
But I guess that if you merged those 2 PRs (well actually this one would have to be modified first if #2148 is merged but I'll do it then), many use cases would be covered and people should stop asking for more support :) (except maybe for removing the static modifier but this would be wrong IMO and prevent entirely this PR).
---------------------------------------------------------------------------
by fabpot at 2011/09/28 11:47:10 -0700
@jalliot: #2148 has been merged. Can you update this PR accordingly? thanks.
---------------------------------------------------------------------------
by jalliot at 2011/09/28 12:00:44 -0700
Sure thing. Will do it as well as removing the check for the interface tonight or tomorrow :)
---------------------------------------------------------------------------
by jalliot at 2011/09/29 08:53:17 -0700
@fabpot Check for interface removed and #2148 merged. Also rebased on latest master.
---------------------------------------------------------------------------
by fabpot at 2011/09/29 09:09:11 -0700
Tests do not pass.
---------------------------------------------------------------------------
by jalliot at 2011/09/29 09:18:48 -0700
@fabpot Fixed
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)
* 2.0:
[Validator] added support for grapheme_strlen when mbstring is not installed but intl is installed
removed separator of choice widget when the separator is null
Commits
-------
dd20f01 Fixed assets:install to use a relative path instead of an absolute
Discussion
----------
[2.1] Fixed "assets:install" to create relative instead of absolute symlinks
This is a fairly simple fix so that the symlinks are relative to the resources rather than an absolute path that breaks from machine-to-machine or upon deployment.
We were trying to figure out why styles were messed up for other contributors, but then found that the paths were hard-coded for my machine :)
---------------------------------------------------------------------------
by ericclemmons at 2011/06/04 09:44:11 -0700
Any other thoughts/updates on this?
---------------------------------------------------------------------------
by fabpot at 2011/06/04 22:31:39 -0700
We have such a feature in symfony1 and IIRC it does not work very well. One problem is when you use symlink for some bundles. Then, you should not use a relative symlink as there is a common path between the two.
---------------------------------------------------------------------------
by ericclemmons at 2011/06/05 09:55:00 -0700
Sorry, I didn't think that we would be an issue since the Bundle "Best Practices" states to not include other bundles as dependencies.
If absolute links are a must, then the next alternative is for collaborators to add "/web/bundles" to .gitignore and each person run "assets:install" upon installation/update.
I was personally hoping there were a way to have this versioned for easier deployment.
On Jun 5, 2011, at 12:31 AM, fabpot<reply@reply.github.com> wrote:
> We have such a feature in symfony1 and IIRC it does not work very well. One problem is when you use symlink for some bundles. Then, you should not use a relative symlink as there is a common path between the two.
>
> --
> Reply to this email directly or view it on GitHub:
> https://github.com/symfony/symfony/pull/1173#issuecomment-1303600
---------------------------------------------------------------------------
by henrikbjorn at 2011/06/27 04:56:58 -0700
``` php
<?php
// ...
->addOption('relative', null, InputOption::VALUE_NONE, 'The --symlink option will generate relative paths')
// ...
```
and just default to absolute paths ?
---------------------------------------------------------------------------
by ericclemmons at 2011/06/27 08:37:50 -0700
I'm very supportive of that compromise. Up to @fabpot if I should add this back in, since relative paths were apparently problematic with symfony1.
---------------------------------------------------------------------------
by sbusch at 2011/07/15 08:46:01 -0700
+1
I'm developing on Mac and the files are mounted on a Linux box which serves the project. The paths are not the same on those two systems. If I accidentally install assets on my Mac the absolute paths won't work on the Linux webserver.
Other scenario: one teammate could add those symlinks by accident to the git repository, which breaks all other installations.
Relative symlinks could help a lot in these cases.
---------------------------------------------------------------------------
by ericclemmons at 2011/07/15 08:47:53 -0700
@sbusch Your issues are the same as mine, which prompted this ticket :)
Until this gets @fabpot's blessing, it's best to simply add `web/bundles` to your `.gitignore` file and tell your users to always run `assets:install --symlink` each time they pull down code & something breaks ;)
---------------------------------------------------------------------------
by sbusch at 2011/07/15 08:58:33 -0700
The handling (calculation) of relative symlinks IMO fits better to the `symlink()` method of `\Symfony\Component\HttpKernel\Util\Filesystem`. Possible method signature:
symfony/src/Symfony/Component/HttpKernel/Util/Filesystem.php:
```php
<?php
// ...
/**
* Creates a symbolic link or copy a directory.
*
* @param string $originDir The origin directory path
* @param string $targetDir The symbolic link name
* @param Boolean $copyOnWindows Whether to copy files if on Windows
* @param Boolean $makeRelative Whether to try to create a relative link
*/
public function symlink($originDir, $targetDir, $copyOnWindows = false, $makeRelative = false)
{
```
And what about changing the `--symlink` option to optionally have a value, instead of adding a new depending option? E.g. `--symlink[=absolute|relative]`, with "absolute" as default:
symfony/src/Symfony/Bundle/FrameworkBundle/Command/AssetsInstallCommand.php:
```php
<?php
// ...
class AssetsInstallCommand extends ContainerAwareCommand
{
/**
* @see Command
*/
protected function configure()
{
$this
->setDefinition(array(
new InputArgument('target', InputArgument::REQUIRED, 'The target directory (usually "web")'),
))
->addOption('symlink', null, InputOption::VALUE_OPTIONAL, 'Symlinks the assets instead of copying it. Allowed values: "absolute" (default) and "relative".', 'absolute')
->setHelp(<<<EOT
The <info>assets:install</info> command installs bundle assets into a given
directory (e.g. the web directory).
<info>./app/console assets:install web [--symlink]</info>
A "bundles" directory will be created inside the target directory, and the
"Resources/public" directory of each bundle will be copied into it.
To create a symlink to each bundle instead of copying its assets, use the
<info>--symlink</info> option. Use <info>--symlink=relative</info> for relative symlinks.
EOT
)
->setName('assets:install')
;
}
/**
* @see Command
*
* @throws \InvalidArgumentException When the target directory does not exist
*/
protected function execute(InputInterface $input, OutputInterface $output)
{
if (!is_dir($input->getArgument('target'))) {
throw new \InvalidArgumentException(sprintf('The target directory "%s" does not exist.', $input->getArgument('target')));
}
if ($input->hasOption('symlink'))
{
if (!function_exists('symlink')) {
throw new \InvalidArgumentException('The symlink() function is not available on your system. You need to install the assets without the --symlink option.');
}
if (!in_array($input->getOption('symlink'), array('absolute', 'relative'))) {
throw new \InvalidArgumentException(sprintf('Invalid value "%s" for option "symlink"', $input->getOption('symlink')));
}
}
$filesystem = $this->getContainer()->get('filesystem');
// Create the bundles directory otherwise symlink will fail.
$filesystem->mkdir($input->getArgument('target').'/bundles/', 0777);
foreach ($this->getContainer()->get('kernel')->getBundles() as $bundle) {
$originDir = $bundle->getPath().'/Resources/public';
if (is_dir($originDir)) {
$targetDir = $input->getArgument('target').'/bundles/'.preg_replace('/bundle$/', '', strtolower($bundle->getName()));
$output->writeln(sprintf('Installing assets for <comment>%s</comment> into <comment>%s</comment>', $bundle->getNamespace(), $targetDir));
$filesystem->remove($targetDir);
if ($input->hasOption('symlink')) {
$filesystem->symlink($originDir, $targetDir, false, $input->getOption('symlink') == 'relative');
} else {
$filesystem->mkdir($targetDir, 0777);
$filesystem->mirror($originDir, $targetDir);
}
}
}
}
```
---------------------------------------------------------------------------
by sbusch at 2011/07/15 09:04:46 -0700
@ericclemmons: yes, that's our current workaround. I started with manually converting absolute links to relative ones, but that quickly got very annoying ;-)
After that I tried to implement the generation of relative links by myself (where the proposals of my previous comment come from) until I found your PR.
---------------------------------------------------------------------------
by henrikbjorn at 2011/07/18 00:20:38 -0700
@sbush if it defaults to something how would you turn it off ?
---------------------------------------------------------------------------
by stof at 2011/07/18 00:26:16 -0700
@henrikbjorn a default value for the option is used when using ``--symlink`` without the value. If you don't use the option at all, it is disabled.
---------------------------------------------------------------------------
by stof at 2011/07/18 11:58:29 -0700
In fact no. The default value seems to be also used when the option is not set at all. @fabpot is this intended ?
---------------------------------------------------------------------------
by Seldaek at 2011/07/19 05:18:29 -0700
Symlinks on windows, although technically possible, don't quite work with PHP on most setups. Also git doesn't seem to support them either on windows (not sure why not). For those reasons, and although I'm sure this doesn't apply to every project, I would recommend you just have everyone run `assets:install [--symlink]` on their local machine, and make that command run on the server as part of your deployment process.
---------------------------------------------------------------------------
by ericclemmons at 2011/07/19 06:15:34 -0700
Nobody is even entertaining --relative?
On Jul 19, 2011, at 7:18 AM, Seldaek<reply@reply.github.com> wrote:
> Symlinks on windows, although technically possible, don't quite work with PHP on most setups. Also git doesn't seem to support them either on windows (not sure why not). For those reasons, and although I'm sure this doesn't apply to every project, I would recommend you just have everyone run `assets:install [--symlink]` on their local machine, and make that command run on the server as part of your deployment process.
>
> --
> Reply to this email directly or view it on GitHub:
> https://github.com/symfony/symfony/pull/1173#issuecomment-1606463
---------------------------------------------------------------------------
by Gregwar at 2011/08/10 08:56:27 -0700
I agree with the idea of proposing a --relative option, I'm currently working on a Samba mounted filesystem and I'm forced to create manually symlinks to get things working since the paths are not the same
Commits
-------
f7bf7b5 fixed condition
181332b added a Controller:getUser() shortcut to recover the current user
Discussion
----------
[2.1] added a Controller:getUser() shortcut to recover the current user
Commits
-------
d675c28 [FrameworkBundle] Use Router instead of RouterInterface
ae7ae8d [FrameworkBundle] Moved router_listener from web to router.xml since it depends on the router
35a9023 [FrameworkBundle] Added isEnabled to Router commands, fixes#1467536d979 [Console] Added Command::isEnabled method that defines whether to add the command or not
Discussion
----------
[2.1] [Console] Added Command::isEnabled method
This addresses #1467.
The idea is to allow commands to evaluate whether they can run or not, since they are automatically registered.
- It's useful for the two router:* commands since they're optional (router can be disabled), but part of the FrameworkBundle that is not really optional.
- It could be useful for third party code as well.
- It's BC.
- aa95bb0d395810b29a3e654673e130736d9d1080 should address the issue in #1467, while the other commits just make sure the command is not registered at all if the router isn't standard.
One issue remains though:
- A few other services like twig helpers get the `ròuter` injected, this means that if there is really **no** router service defined, there is still an error. I'm not sure how to fix those beyond adding `on-invalid="null"` but I'm not sure if that's desirable. I guess we could argue that the router is a big candidate for replacement/suppression, and as such it should be truly optional, but if we do it I don't know where it'll lead. I don't want to end up in a situation where half the dependencies are optional to support every possible combination. @fabpot wdyt?
---------------------------------------------------------------------------
by kriswallsmith at 2011/06/28 16:19:46 -0700
I'd rather see us not register a command instead of register and then disable it. Can we do the same thing you've done here in the bundle's registerCommands() method?
---------------------------------------------------------------------------
by Seldaek at 2011/06/28 16:51:36 -0700
Note that it's never really registered. During the registration it's checked and skipped if not enabled.
However, doing it as you suggest means overriding/copy-pasting all the code from the core Bundle class, which I don't like so much. It also means adding code specific to those two commands in a somewhat unrelated place, which I also don't like.
I'm not saying the current solution is perfect, but from the alternatives I considered, it's the best I have found.
---------------------------------------------------------------------------
by stof at 2011/09/04 04:58:04 -0700
@Seldaek your branch conflicts with master. could you rebase it ?
@fabpot what do you think about this PR ?
---------------------------------------------------------------------------
by Seldaek at 2011/09/04 08:39:05 -0700
Rebased
Builds upon aead4a9836180cabae4d47fe27c634dcd79ac8f2, which prematurely removed request scoping from the assets templating helper in all cases. The helper need only be request-scoped if one or more request-scoped packages (e.g. PathPackages) are injected into it. This change makes it possible to utilize the assets helper outside of a request (e.g. during a console script).
To ensure that the assets helper is not assigned a request scope, all asset base URL's must be defined for all packages (default and any named) and both protocols: HTTP and SSL. The included test config fixtures concisely accomplish this by specifying a single HTTPS URL as the base URL for our default and named package, since FrameworkExtension's Configuration conveniently registers this URL for both protocols.
No other helpers have request scope and the assets helper's parameters don't appear to depend on the request in any way, so this appears to be unnecessary. As-is, request scope here prevents use of the assets helper from a console command that may need to internally render a template.
Commits
-------
afc0971 make it easier to customize the cache lookup in the TemplateLocator
Discussion
----------
make it easier to customize the cache lookup in the TemplateLocator
---------------------------------------------------------------------------
by fabpot at 2011/09/22 01:21:56 -0700
Do you any use case for that?
---------------------------------------------------------------------------
by lsmith77 at 2011/09/22 01:27:06 -0700
Yes in the ThemeBundle we override the default cache to include the key + theme:
https://github.com/liip/liipthemebundle/pull/9/files#L7R53
---------------------------------------------------------------------------
by lsmith77 at 2011/09/22 01:29:18 -0700
though i just thought about it .. if we do not put this into 2.0, it might not be worth the trouble.
---------------------------------------------------------------------------
by pjedrzejewski at 2011/09/22 02:19:42 -0700
+1 for this.
Commits
-------
83199ae [FrameworkBundle] Fix unintuitive merging behavior for assets_base_urls
4061114 [FrameworkBundle] fixes unintuitive merging behavior
Discussion
----------
[FrameworkBundle] fixes unintuitive merging behavior
---------------------------------------------------------------------------
by fabpot at 2011/09/16 10:04:53 -0700
I think this is a "bug", no? If this is the case, then we need to fix the 2.0 branch.
---------------------------------------------------------------------------
by schmittjoh at 2011/09/17 00:34:14 -0700
It is a change in behavior, but whether or how you want to merge this is really up to you.
---------------------------------------------------------------------------
by jmikola at 2011/09/19 08:09:26 -0700
I was about to create a PR for this very same quirk, as this was causing my CDN's for various environments to all get merged together. I think we can get away with merging it directly to 2.0 since `assets_base_urls` are hardly covered in the documentation at all.
Once this gets merged, I wouldn't mind writing up a blurb on them to explain how the shorthand syntax works and this merging strategy.
---------------------------------------------------------------------------
by jmikola at 2011/09/19 08:28:21 -0700
I just noticed this PR only fixes the `base_urls` config option under `packages`. We should also correct this behavior for `assets_base_urls`, which appears further up in FrameworkBundle's Configuration.php file.
---------------------------------------------------------------------------
by jmikola at 2011/09/19 08:44:57 -0700
@schmittjoh: I have the second commit for this sitting in https://github.com/jmikola/symfony/tree/configFix (rebased on your branch) if you'd prefer to merge that into your branch to update this PR.
---------------------------------------------------------------------------
by schmittjoh at 2011/09/19 08:55:42 -0700
Merged it in. Thanks.
---------------------------------------------------------------------------
by fabpot at 2011/09/19 09:01:27 -0700
ok, I'm going to merge this into master.
@jmikola: Can you submit documentation for the new way?
Commits
-------
b9b6084 [FrameworkBundle] Updated Czech translations for image validator
Discussion
----------
[FrameworkBundle] Updated Czech translations for image validator
---------------------------------------------------------------------------
by stof at 2011/09/10 12:10:08 -0700
Could you send a PR against the 2.0 branch for the trans-unit up to 41 which are already part of 2.0 ? Adding 42 to 46 should then still be done in master as they are new for 2.1
Commits
-------
9fe87be More explicit default value for assets_version_format
Discussion
----------
Fixed default asset version format
This is not needed as it is already the value that is set when null in ``Symfony\Component\Templating\Asset\Package`` but that would make it clearer for people who just read the configuration.
