This PR was squashed before being merged into the 2.3 branch (closes#14842).
Discussion
----------
[Security][bugfix] "Remember me" cookie cleared on logout with custom "secure"/"httponly" config options [1]
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #14822
| License | MIT
| Doc PR | ~
* test now always pass "secure" and "httponly" options, as they are required
* could be considered BC, but [`RememberMeFactory` passes them](https://github.com/symfony/symfony/blob/2.3/src/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/RememberMeFactory.php#L21), so they should've always been treated as required
* I can squash the commits before merging
* Alternative solution: #14843
Commits
-------
18b1c6a [Security][bugfix] "Remember me" cookie cleared on logout with custom "secure"/"httponly" config options [1]
The `SwitchUserEvent` is triggered in case an account is switched. This works okay while switching to the user, but on exit the `SwitchUserEvent` is triggered again with the original User. That User was not initialized by the provider yet.
load user by UserInterface instead of username
This PR was squashed before being merged into the 2.3 branch (closes#14670).
Discussion
----------
[Security] TokenBasedRememberMeServices test to show why encoding username is required
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #14577
| License | MIT
| Doc PR | no
241538d shows that it's not actually tested, 257b796 reimplements it with test.
I can remove the POC commit if it's not needed.
Commits
-------
63a9736 [Security] TokenBasedRememberMeServices test to show why encoding username is required
This PR was squashed before being merged into the 2.3 branch (closes#14678).
Discussion
----------
[Security] AbstractRememberMeServices::encodeCookie() validates cookie parts
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #14577
| License | MIT
| Doc PR | no
`AbstractRememberMeServices::encodeCookie()` guards against `COOKIE_DELIMITER` in `$cookieParts`.
* it would make `AbstractRememberMeServices::cookieDecode()` broken
* all current extending classes do it anyway (see #14670 )
* added tests – it's not a public method, but it is expected to be used by user implementations – as such, it's good to know that it works properly
Commits
-------
464c39a [Security] AbstractRememberMeServices::encodeCookie() validates cookie parts
This PR was merged into the 2.3 branch.
Discussion
----------
CS: Pre incrementation/decrementation should be used if possible
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
Fixes provided by new fixer: https://github.com/FriendsOfPHP/PHP-CS-Fixer/pull/1113
If this pr is merged I would change the level of the fixer to `symfony`.
Commits
-------
c5123d6 CS: Pre incrementation/decrementation should be used if possible
This PR was merged into the 2.3 branch.
Discussion
----------
CS: fix some license headers
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | ?
| Fixed tickets | N/A
| License | MIT
| Doc PR | N/A
Commits
-------
2b74841 CS: fix some license headers
Reduce couple count calls in [Yaml]
Modernize type casting, fix several strict comparisons
Unsets merged
Elvis operator usage
Short syntax for applied operations
This PR was squashed before being merged into the 2.3 branch (closes#13466).
Discussion
----------
[Security] Remove ContextListener's onKernelResponse listener as it is used
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
The context listeners are specific to a particular firewall, and as such, should not be applied if the current request doesn't match that context listener. To avoid this, the context listener can remove itself from the dispatcher as it is called.
This comes in to affect when two or more firewalls are setup and using the same kernel for multiple requests. Assuming there are two firewalls 'site' and 'admin'
- Request comes in matching 'site' firewall, 'site' ContextListener adds it's onKernelResponse method to the dispatcher
- Succesful auth for 'site'
- ContextListener writes token to session
- Request comes in matching 'admin' firewall, 'admin' ContextListener can't find anything in the session, so nulls the token in the security context
- 'site' ContextListener listens for response, can't find a token in the security context so removes the 'site' token from the session
Commits
-------
380d805 [Security] Remove ContextListener's onKernelResponse listener as it is used
This PR was merged into the 2.3 branch.
Discussion
----------
unified return null usages
| Q | A
| ------------- | ---
| License | MIT
This PR unifies the way we return `null` from a function or method:
* always use `return;` instead of `return null;` (the current code base uses both);
* never use `return;` at the end of a function/method.
Commits
-------
d1d569b unified return null usages