Parse the _controller format in sub-requests #23013
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
weaverryan
force-pushed
the
normalize-controller-format
branch
from
e36a036 to
89c6891
Compare
Contributor
|
Another solution would be to resolve the controller name with the This would remove the BC break. |
stof
requested changes
Jun 1, 2017
| <tag name="kernel.event_subscriber" /> | ||
| </service> | ||
|
|
||
| <service id="resolve_controller_name_subscriber" class="Symfony\Bundle\FrameworkBundle\EventListener\ResolveControllerNameSubscriber" public="true"> |
Member
There was a problem hiding this comment.
should not be public. Listeners can be private now, and there is no BC issue about making this one private as it is new
| public static function getSubscribedEvents() | ||
| { | ||
| return array( | ||
| KernelEvents::REQUEST => array(array('onKernelRequest', 24)), |
Member
There was a problem hiding this comment.
you can simplify this as array('onKernelRequest', 24)
weaverryan
force-pushed
the
normalize-controller-format
branch
2 times, most recently
from
8530c9e to
3856932
Compare
weaverryan
force-pushed
the
normalize-controller-format
branch
from
3856932 to
9578fd3
Compare
Member
Author
|
@jvasseur I really like that idea... but the In the mean time, I've made updates from Stof's comments! |
Member
|
Thank you @weaverryan. |
fabpot
added a commit
that referenced
this pull request
Jun 2, 2017
This PR was merged into the 3.3 branch. Discussion ---------- Parse the _controller format in sub-requests | Q | A | ------------- | --- | Branch? | 3.3 | Bug fix? | yes | New feature? | no | BC breaks? | possibly (edge case) | Deprecations? | no | Tests pass? | yes | Fixed tickets | #22966 | License | MIT | Doc PR | n/a As mentioned on the issue (#22966 (comment)), the new "controller service args" functionality relies on the `_controller` attribute to be in either the service format `App\Controller\Foo:bar` or at least the final parsed format `App\Controller\Foo::bar`. But when you make a sub-request with the `App:Foo:bar` format, the `ControllerResolver` correctly parses this, but the `_controller` request attribute will always contain the original `App:Foo:bar` format. That causes the `ServiceValueResolver` to fail. The only way I can think to fix this - reliably - is to parse the `_controller` attribute in a listener. And this, works great! Notes: A) There is a small chance for a BC break: if you were relying on the `_controller` old format in a `kernel.request` format in the framework, in a listener between the priority of 25 and 31 for sub-requests (because normal requests have `_controller` normalized during routing)... then you will see a behavior change. B) We could load the `ControllerNameParser` lazily via a service locator. C) We could deprecate calling the parser in the FW's `ControllerResolver`. Along with (B), I think it would (in 4.0) mean that the `ControllerNameParser` is not instantiated at runtime (except for sub-requests). If someone can think of a different/better solution, please let me know! Cheers! Commits ------- 9578fd3 Adding a new event subscriber that "parses" the _controller attribute in the FW
Merged
1 task
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As mentioned on the issue (#22966 (comment)), the new "controller service args" functionality relies on the
_controllerattribute to be in either the service formatApp\Controller\Foo:baror at least the final parsed formatApp\Controller\Foo::bar. But when you make a sub-request with theApp:Foo:barformat, theControllerResolvercorrectly parses this, but the_controllerrequest attribute will always contain the originalApp:Foo:barformat. That causes theServiceValueResolverto fail.The only way I can think to fix this - reliably - is to parse the
_controllerattribute in a listener. And this, works great! Notes:A) There is a small chance for a BC break: if you were relying on the
_controllerold format in akernel.requestformat in the framework, in a listener between the priority of 25 and 31 for sub-requests (because normal requests have_controllernormalized during routing)... then you will see a behavior change.B) We could load the
ControllerNameParserlazily via a service locator.C) We could deprecate calling the parser in the FW's
ControllerResolver. Along with (B), I think it would (in 4.0) mean that theControllerNameParseris not instantiated at runtime (except for sub-requests).If someone can think of a different/better solution, please let me know!
Cheers!