This PR was merged into the 2.7 branch.
Discussion
----------
[DependencyInjection] move tags from decorated to decorating service
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20150
| License | MIT
| Doc PR |
Commits
-------
7def83c move tags from decorated to decorating service
This PR was merged into the 2.7 branch.
Discussion
----------
[HttpCache] fix: do not cache OPTIONS request
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | https://github.com/symfony/symfony/issues/20202
| License | MIT
| Doc PR | -
The HttpCache should not cache any OPTIONS request as they are by spec not cacheable (mentioned here https://github.com/symfony/symfony/issues/20202#issuecomment-253033144 by @xabbuh).
Commits
-------
c43de7f [HttpCache] fix: do not cache OPTIONS request
This PR was submitted for the master branch but it was merged into the 2.7 branch instead (closes#20146).
Discussion
----------
[Validator] Prevent infinite loop in PropertyMetadata
| Q | A
| ------------- | ---
| Branch? | 3.1
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |n/a
| License | MIT
| Doc PR | n/a
This commit fixes a possible infinite loop in PropertyMetadata when the PropertyMetadata class was constructed with an existing property of class and later used (after being serialized and cached) on that same class while that property no longer existing. `get_parent_class` will return false when there is no parent class and `property_existing` will then keeping return false causing the `while` loop to be infinite.
Commits
-------
c1ae7b6 Prevent infinite loop in PropertyMetadata
This PR was merged into the 2.7 branch.
Discussion
----------
[travis] Test on PHP 7.1
| Q | A
| ------------- | ---
| Branch? | 2.7
| Tests pass? | let's see
| License | MIT
PHP 7.1 RC2 has been released today.
Commits
-------
6602e21 [travis] Test on PHP 7.1
This PR was merged into the 2.7 branch.
Discussion
----------
[FrameworkBundle] Convert null prefix to an empty string in translation:update
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20044
| License | MIT
| Doc PR | n/a
This command needs the ability to use an empty string as prefix, which is not possible using `bin/console translation:update --prefix=""` because `$argv` doesn't parse empty strings thus the value is converted to `null` by `ArgvInput` (only since #19946, before the option was not considered to be set, giving the default `'__'` thus this should be fine from a BC pov).
Here I propose to explicitly convert the `prefix` value to an empty string if set to `null`, as it is a very specific need and we can't guess that from `ArgvInput`.
An other way to fix it could be to add a `--no-prefix` option to the command but I don't think it is worth it, and it couldn't be treated as a bug fix thus not fixed before `3.2`.
Commits
-------
f02b687 [FrameworkBundle] Convert null prefix to an empty string in translation:update command
This PR was merged into the 2.7 branch.
Discussion
----------
[Security] $attributes can be anything, but RoleVoter assumes strings
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | yes
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #18042
| License | MIT
| Doc PR | reference to the documentation PR, if any
Commits
-------
ad3ac95 bug #18042 [Security] $attributes can be anything, but RoleVoter assumes strings
This PR was submitted for the master branch but it was merged into the 2.7 branch instead (closes#20127).
Discussion
----------
[HttpFoundation] JSONP callback validation
| Q | A
| ------------- | ---
| Branch? | "master"
| Bug fix? | yes
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #17923
| License | MIT
| Doc PR | reference to the documentation PR, if any
Maybe this is too small for a new dep, but at least it's stable. Symfony itself will make no assumption on validation by default, ie. things should keep working as usual.
Commits
-------
1159f8b [HttpFoundation] JSONP callback validation
This PR was squashed before being merged into the 2.7 branch (closes#20159).
Discussion
----------
[Console] Improved the explanation of the hasOption() method
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #17597
| License | MIT
| Doc PR | -
Commits
-------
29226ef [Console] Improved the explanation of the hasOption() method
This PR was merged into the 2.7 branch.
Discussion
----------
[Console] Escape question text and default value in SymfonyStyle::ask()
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #19493
| License | MIT
| Doc PR | n/a
Commits
-------
eed3cc5 [Console] Escape default value and question in SymfonyStyle::ask()
This PR was merged into the 2.7 branch.
Discussion
----------
[Console] Fix validation of empty values using SymfonyQuestionHelper::ask()
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? |no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
When using `QuestionHelper::ask()` it's allowed to return an empty value as answer, e.g:
```php
$helper = new QuestionHelper();
$question = new Question('foo', false);
$question->setValidator(function ($v) { return $v; });
$answer = $helper->ask($input, $output, $question);
```
Just typing `enter` for answering this question works, the value of `$answer` would be `false`.
But doing the same with `SymfonyQuestionHelper::ask()`:
```php
$helper = new SymfonyQuestionHelper();
$question = new Question('foo', false);
$question->setValidator(function ($v) { return $v; });
$answer = $helper->ask($input, $output, $question);
```
> [ERROR] A value is required.
Same for `''` or `null`.
Here I kept the same check but used as default validator, if a validator is set and allows an empty value to be returned then it's ok.
Also I am not sure about if this default validator should be kept, imho we should be consistent with the default question helper, using the `SymfonyQuestionHelper` should only impact the output.
Diff best viewed [like this](https://github.com/symfony/symfony/pull/20141/files?w=1)
ping @kbond
Commits
-------
a8b910b [Console] Fix validation of null values using SymfonyStyle::ask()
This PR was merged into the 2.7 branch.
Discussion
----------
[FrameworkBundle] Alter container class instead of kernel name in cache:clear command
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #19912
| License | MIT
| Doc PR | -
The decision to alter kernel's name instead of the container's class dates from cc3a40ed78
But this is causing issues such as #19912. Looking more carefully, the real intent is to change the container class and there is no need to alter also the kernel's name at this stage.
Commits
-------
73c9693 [FrameworkBundle] Alter container class instead of kernel name in cache:clear command
This PR was merged into the 2.7 branch.
Discussion
----------
[VarDumper] Fix ReflectionNamedType->getName() detection
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
There is no side effect to this patch, except that __toString were still used, whereas getName should have been (both methods return the same, but __toString is deprecated since 7.1).
I somehow missed that getName was on a new `ReflectionNamedType` class...
Commits
-------
dda18df [VarDumper] Fix ReflectionNamedType->getName() detection
This PR was merged into the 2.7 branch.
Discussion
----------
[travis] Use hhvm-stable instead of discontinued hhvm-3.12
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Commits
-------
2ac1586 [travis] Use hhvm-stable instead of discontinued hhvm-3.12
This PR was squashed before being merged into the 2.7 branch (closes#20102).
Discussion
----------
[Validator] Url validator not validating hosts ending in a number
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no (Should validate slightly more urls)
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | None that I could find
| License | MIT
| Doc PR |
We ran into an issue where the URL validator wasn't validating URL that we got from a legitimate (intranet) URL. Afaict from skimming RFCs the URL `http://myhost123/` is valid, however the current regular expression doesn't accept the last part of a hostname (in this case the entire hostname is the last part as there are no periods in the host).
I've fixed this by adding the numbers unicode character property to the regular expression, I'm not sure but symbol characters (`\pS`) might need to be added to the group as well.
Commits
-------
934c434 [Validator] Url validator not validating hosts ending in a number
This PR was squashed before being merged into the 2.7 branch (closes#20132).
Discussion
----------
Use "more entropy" option for uniqid()
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
As you can see in [the source code of the PHP uniqid() function](1c295d4a9a/ext/standard/uniqid.c (L68)) when you don't pass `true` as the second argument, PHP sleeps the application for 1 microsecond (`usleep(1)`)
Symfony uses `true` almost everywhere, but there are still some places (mostly in bundles) where this is not the case.
Commits
-------
4403e28 Use "more entropy" option for uniqid()
This PR was merged into the 2.7 branch.
Discussion
----------
[Validator] Reset constraint options
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20106
| License | MIT
| Doc PR | reference to the documentation PR, if any
I think it's good to reset the internal pointer of the options array if we depend on `key($options)`. I've added tests to clarify we intentionally check for the first key.
The current behavior actually differs for hhvm/php7 - https://3v4l.org/e6r6E
Commits
-------
4c6ddd4 reset constraint options
This PR was submitted for the master branch but it was merged into the 2.7 branch instead (closes#20116).
Discussion
----------
fixed AddConstraintValidatorsPass config
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
Commits
-------
996c1ee added checks for public services on compiler passes that use service id and not references
This PR was merged into the 2.7 branch.
Discussion
----------
Fix#19943 Make sure to process each interface metadata only once
| Q | A
| ------------- | ---
| Branch? | 2.7+
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #19943
| License | MIT
| Doc PR | N/A
Here is the fix for #19943, If you have `InterfaceA <- InterfaceB <- Class` with a constraint on a method defined on `InterfaceA`, the constraint and its eventual violations are currently validated and reported twice.
Copy from https://github.com/symfony/symfony/issues/19943#issuecomment-250238529:
As far as I can see, the problem seems to arise in [`LazyLoadingMetadataFactory`](https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Validator/Mapping/Factory/LazyLoadingMetadataFactory.php#L117-L123):
[`ReflectionClass::getInterfaces()`](http://php.net/manual/en/reflectionclass.getinterfaces.php) returns both interfaces implemented directly and through inheritance (either through another interface or through a parent class). In the end, the following process occurs:
1. `PriceInterface` is parsed and its `NotBlank` constraint on `value` is loaded
2. `VariablePriceInterface` is parsed and inherits `PriceInterface`'s constraints (which is OK).
3. `ProductPrice` is parsed and inherits both `PriceInterface` and `VariablePriceInterface`'s constraints, which leads to a duplicated `NotBlank` constraint, one from each Interface.
The Best Way ™️ would be to be able to extract the list of interfaces implemented by a class directly only. However, the process seems a bit intricate... I will start working on it and prepare a PR to that effect. However, if any of you has a better idea, I'm all hears...
TODO:
- [x] Regression tests to make sure the bug doesn't reappear
Commits
-------
2f9b65aFix#19943 Make sure to process each interface metadata only once
This PR was merged into the 2.7 branch.
Discussion
----------
[TwigBundle] Fix CacheWarmingTest are order dependent
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Fix issue reported in #20065
Commits
-------
fcd6ec2 [TwigBundle] Fix CacheWarmingTest are order dependent