This PR was merged into the 2.3 branch.
Discussion
----------
use stable 1.0.x Composer versions
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
This will make our tests pass on AppVeyor again where we always used the latest Composer snapshot (this is 1.1.x-dev now which the `hirak/prestissimo` plugin is not compatible yet, but I opened hirak/prestissimo#79 to make that happen).
Commits
-------
10e2dfb use stable 1.0.x Composer versions
This PR was merged into the 2.3 branch.
Discussion
----------
[PropertyAccess] ->getValue() should be read-only
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #18601
| License | MIT
| Doc PR | -
Commits
-------
fa68529 [PropertyAccess] ->getValue() should be read-only
This PR was merged into the 2.3 branch.
Discussion
----------
[Routing] add query param if value is different from default
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | yes
| BC breaks? | most likely not
| Deprecations? | no
| Tests pass? | no
| Fixed tickets | #10940, #18111, #18035
| License | MIT
| Doc PR | -
Commits
-------
1ef2edf [Routing] add query param if value is different from default
This PR was merged into the 2.3 branch.
Discussion
----------
[DependencyInjection] Remove check for YAML class in CrossCheckTest
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
I suppose it's a leftover from when YAML component only was added in suggest.
Commits
-------
af92275 [DependencyInjection] Remove YAML check in CrossCheckTest
This PR was submitted for the master branch but it was merged into the 2.3 branch instead (closes#18515).
Discussion
----------
[Filesystem] Better error handling in remove()
| Q | A
| ------------- | ---
| Branch? | master (to be moved on 2.3 when merging)
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #18512
| License | MIT
| Doc PR | -
Commits
-------
b848ddb [Filesystem] Better error handling in remove()
This PR was merged into the 2.3 branch.
Discussion
----------
[DependencyInjection] Add coverage for all invalid arguments in exportParameters
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
Commits
-------
87c4f23 [DependencyInjection] Add coverage for all invalid arguments in exportParameters
This PR was merged into the 2.3 branch.
Discussion
----------
[DomCrawler] Fix select option with empty value
| Q | A
| ------------- | ---
| Branch? | 2.3+
| Bug fix? | yes
| New feature? | no
| BC breaks? | no?
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
While using the Laravel's testing suite which makes use of the Symfony DOM Crawler (v3.0.2). I have been populating a form with a select which has a value which can be an empty value.
For example, with this select you can choose your gender or leave it empty if you don't want to specify:
```html
<select name="gender">
<option selected></option>
<option>Female</option>
<option>Male</option>
</select>
```
When the `DomCrawler\Field::getValue()` is called I was expect to get the value `''` however I was actually getting `'on'`. This is caused by the [DomCrawler\Field::buildOptionValue()](https://github.com/symfony/symfony/blob/master/src/Symfony/Component/DomCrawler/Field/ChoiceFormField.php#L262-L271) sets the default value to 'on' when there is no value which makes sense for ratios and checkboxes but not for select.
I have tracked this bug back to v2.3 but it is still present in v3, however, the default value was changed from '1' to 'on' in v2.5 which means that this patch will conflict when merging up the maintained versions.
Commits
-------
58276a2 Fix Dom Crawler select option with empty value
This PR was merged into the 2.3 branch.
Discussion
----------
[Form] Remove unnecessary option assignment
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Option assignment is not required because the prototype inherits this option from the parent form via standard inheritance mechanism.
Related pull requests: #16959, #18317
Commits
-------
da8a197 Remove unnecessary option assignment
This PR was merged into the 2.3 branch.
Discussion
----------
[Security] Fixed SwitchUserListener when exiting an impersonation with AnonymousToken
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
If you configure a firewall with switch user with `role: IS_AUTHENTICATED_ANONYMOUSLY` it's impossible to exit the
impersonation because the next line `$this->provider->refreshUser($original->getUser())` will fail. It fails because `RefreshUser`
expects an instance of `UserInterface` and here it's a string.
Therefore, it does not make sense to refresh an Anonymous Token, right ?
Commits
-------
59fea72 [Security] Fixed SwitchUserListener when exiting an impersonication with AnonymousToken
This PR was merged into the 2.3 branch.
Discussion
----------
[Form] fix "prototype" not required when parent form is not required
| Q | A
| ------------- | ---
| Branch? | 2.3+
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #18311
| License | MIT
| Doc PR | ~
Commits
-------
7df9ca2 [Form] fix "prototype" not required when parent form is not required
This PR was submitted for the master branch but it was merged into the 2.3 branch instead (closes#18442).
Discussion
----------
[HttpFoundation] Improve phpdoc
| Q | A
| ------------- | ---
| Branch? | master?
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Improve the phpdoc for the `$default` parameter of the `get()` method. It wasn't clear when the default value would be used (whether the key would not exist or the value was `null` or nullish).
The comment is now in sync with `Symfony\Component\HttpFoundation\ParameterBag::get()`.
Commits
-------
6bfbb2e [HttpFoundation] Improve phpdoc
Improve the phpdoc for the `$default` parameter of the `get()` method. It wasn't clear when the default value would be used (whether the key would not exist or the value was `null` or nullish).
The comment is now in sync with `Symfony\Component\HttpFoundation\ParameterBag::get()`.
This PR was merged into the 2.3 branch.
Discussion
----------
[Logging] Add support for Firefox (43+) in ChromePhpHandler
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | none
| License | MIT
| Doc PR | none
Firefox added support for Chrome Logger a while back in [version 43](https://www.mozilla.org/en-US/firefox/43.0/releasenotes/).
Support for Firefox was added in the ChromePHPHandler of [Monolog 1.18.0](https://github.com/Seldaek/monolog/issues/721), although its not mentioned in the changelog.
This PR is basically just to replace the originally copied regular expression, [with this new one](fa96f6aa8f).
There should be no BC breakage, even though users may be using Monolog versions between 1.11 and 1.18. Monolog's ChromePHPHandler only got minor touch ups and its protocol (and thus output) remained the same.
This fix in fact also adds support for Firefox 43+ to older Monolog's when using the Symfony-version of that handler (which they should be using), since the Symfony-version of ChromePhpHandler completely overwrites the 'headersAccepted'-check of Monolog.
Commits
-------
c8efc4d [Logging] Add support for firefox in ChromePhpHandler
This PR was submitted for the master branch but it was merged into the 2.3 branch instead (closes#18385).
Discussion
----------
Detect CLI color support for Windows 10 build 10586
Newer Windows 10 versions (builds starting from 10586) offer VT100 color support.
See http://www.nivot.org/blog/post/2016/02/04/Windows-10-TH2-(v1511)-Console-Host-Enhancements
| Q | A
| ------------- | ---
| Branch? | master - Maybe it could be backported to other branches too
| Bug fix? | maybe 😉 - Do you you consider a bug not having colors on Windows?
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
Commits
-------
472a7bf Detect CLI color support for Windows 10 build 10586
This PR was merged into the 2.3 branch.
Discussion
----------
[EventDispatcher] Try first if the event is Stopped
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
If you trigger 2 events with the same instance of Event and if a listener in the first dispatch stop the propagation, then the very first listener of the second dispatch is called. It should not IMHO.
Commits
-------
a30e166 [EventDispatcher] Try first if the event is Stopped
If you configure a firewall with switch user with `role: IS_AUTHENTICATED_ANONYMOUSLY` it's impossible to exit the
impersonation because the next line `$this->provider->refreshUser($original->getUser())` will fail. It fails because `RefreshUser`
expects an instance of `UserInterface` and here it's a string.
Therefore, it does not make sense to refresh an Anonymous Token, right ?
This PR was merged into the 2.3 branch.
Discussion
----------
[travis] Disable hirak/prestissimo for deps=low/high tests
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Concurrent composer updates + prestissimo for deps=low/high matrix lines make tests transient.
Lets test each component in its own `$HOME`.
Commits
-------
b8205a8 [travis] Disable hirak/prestissimo for deps=low/high tests
This PR was merged into the 2.3 branch.
Discussion
----------
[ci] Skip dns-sensitive tests when DnsMock is not found
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #18319
| License | MIT
| Doc PR | -
Commits
-------
7e01187 [ci] Skip dns-sensitive tests when DnsMock is not found
This PR was merged into the 2.3 branch.
Discussion
----------
Optimize ReplaceAliasByActualDefinitionPass
| Q | A
| ------------- | ---
| Branch? | 2.3
| Bug fix? | yes (performance)
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Previous implementation passed over every definition for every alias (n*m runtime).
New implementation passes once over all aliases and once over all definitions (n+m).
Also removing needless "restart" logic.
Commits
-------
ab8dc0c Optimize ReplaceAliasByActualDefinitionPass