This PR was merged into the 2.3 branch.
Discussion
----------
[console] Use the description when no help is available
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #15498
| License | MIT
| Doc PR |
Commits
-------
e5d3f25 [console] Use the description when no help is available
This PR was merged into the 2.3 branch.
Discussion
----------
[ci] Run minimal versions on appveyor only
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Hopefully the last round of ci optimizations: what do you think about running the minimal supported version on appveyor only? This removes one matrix line on travis, and gives more flexibility in choosing the minimal version because, where travis only supports a limited list of minimal versions, windows.php.net provides a binary for each and every php ever released.
Commits
-------
e564805 [ci] Run minimal versions on appveyor only
This PR was merged into the 2.3 branch.
Discussion
----------
consistently use str_replace to unify directory separators
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR | -
We used a mix of `str_replace` and `strtr` for the same thing. `str_replace` is faster as [I tested myself](https://3v4l.org/MNrgC) which is logical since strtr adds logic we don't need here.
Commits
-------
b9760ef consistently use str_replace to unify directory separators
This PR was merged into the 2.3 branch.
Discussion
----------
[Security] Add missing docblock in PreAuthenticatedToken
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
I noticed the `PreAuthenticatedToken` does not provide any dockblock about its constructor parameters so PHPStorm uses the `AbstractToken` parameters and think I'm using `PreAuthenticatedToken` badly.
This fix it by adding the missing docblock and I also added the string[] possibility for roles in `AbstractToken` as it's present in the code and the exception message.
Commits
-------
1dda0a6 [Security] Add missing docblock in PreAuthenticatedToken
This PR was merged into the 2.3 branch.
Discussion
----------
Add appveyor.yml for C.I. on Windows
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | YES, both on Travis (Linux) and on Appveyor (Windows)!
| Fixed tickets | #13934, #15049, #14259, #15045, #15444
| License | MIT
| Doc PR | https://github.com/symfony/symfony-docs/pull/5654
- testing two matrix lines:
- one without mbtring nor fileinfo nor intl
- one with these ext enables, intl version 51.2 so that almost no test is skipped on our Intl component
- bug fixes thanks to these harder testing conditions
- some display bug on appveyor, [reported here](http://help.appveyor.com/discussions/suggestions/197-support-ansi-color-codes).
Commits
-------
ea5d656 Windows and Intl fixes
8bbd8d9 Add appveyor.yml for C.I. on Windows
This PR was merged into the 2.3 branch.
Discussion
----------
Implement the support of timezone objects in the stub IntlDateFormatter
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
As of PHP 5.5, the IntlDateFormatter accepts to use DateTimeZone or IntlTimeZone objects as timezone in the constructor (and in the new setTimeZone method) rather than timezone ids.
This is even the proper way to pass a timezone from a DateTime object as DateTimeZone names are not all valid ICU identifiers (and there is a PR on the Twig-extensions repo to use such feature to support things properly: https://github.com/twigphp/Twig-extensions/pull/148).
I'm considering this as a bugfix because it is a mismatch between the stub implementation and the real class.
Note that for simplicity, these objects are accepted on all PHP versions, as reproducing the behavior of older versions is not possible in the stub anyway (triggering a warning and making the instantiating with ``new`` return ``null``). We already have such differences anyway (the ``setTimeZone`` method exists in all PHP versions in the stub)
Commits
-------
2856abe Implement the support of timezone objects in the stub IntlDateFormatter
As of PHP 5.5, the IntlDateFormatter accepts to use DateTimeZone or
IntlTimeZone objects as timezone in the constructor (and in the new
setTimeZone method) rather than timezone ids.
This is even the proper way to pass a timezone from a DateTime object as
DateTimeZone names are not all valid ICU identifiers.
This PR was merged into the 2.3 branch.
Discussion
----------
[Serializer] Add support for variadic arguments in the GetSetNormalizer
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
There were 2 broken cases:
- when the value was passed, the array was passed as argument, becoming the first value of the variadic array. The array needs to be spread into multiple arguments when calling the method
- when the value was missing, the code would throw a ReflectionException, similar to the issue reported in #13690, because a variadic argument is optional but does not have a default value
Commits
-------
704760b Add support for variadic arguments in the GetSetNormalizer
This PR was merged into the 2.3 branch.
Discussion
----------
[Yaml] Nested merge keys
When trying to use nested merge keys in Yaml, the ref sometimes had the wrong value
eg, with:
```yaml
taz: &taz
a: Steve
w:
p: 1234
nested:
<<: *taz
d: Doug
w: &nestedref
p: 12345
z:
<<: *nestedref
```
in this case, the ref `nestedref` had the value "Doug", which cause this error :
Symfony\Component\Yaml\Exception\ParseException: YAML merge keys used with a scalar value instead of an array at line 38 (near "<<: *nestedref").
It now works as expected
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
note: this PR replace #15241
Commits
-------
500c57e [Yaml] Nested merge keys
This PR was squashed before being merged into the 2.3 branch (closes#15263).
Discussion
----------
[HttpFoundation] fixed the check of 'proxy-revalidate' in Response::mustRevalidate()
| Q | A
| ------------- | ---
| Fixed tickets | #15262
| License | MIT
'proxy-revalidate' is not a header on its own but a 'Cache-Control' directive
See http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.9
Commits
-------
6c22f0a [HttpFoundation] fixed the check of 'proxy-revalidate' in Response::mustRevalidate()
This PR was merged into the 2.3 branch.
Discussion
----------
Remove skipping of tests based on ICU data version whenever 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
Many tests being skipped based on the ICU data version don't actually need it. They might be testing code paths not relying on Intl, or not performing assertions on the values depending on the ICU data and so not dependant on the exact ICU version being used.
this is somewhat related to https://github.com/symfony/symfony/issues/14259 as it allows to reduce the number of tests not running on Travis.
Commits
-------
7994513 Remove skipping of tests based on ICU data version whenever possible
This PR was merged into the 2.3 branch.
Discussion
----------
do not dump leading backslashes in class names
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #15001
| License | MIT
| Doc PR |
Commits
-------
ad6cb10 do not dump leading backslashes in class names
This PR was merged into the 2.3 branch.
Discussion
----------
[Config] type specific check for emptiness
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #13736
| License | MIT
| Doc PR |
Commits
-------
0199fbf [Config] type specific check for emptiness
This PR was merged into the 2.3 branch.
Discussion
----------
Fix the handling of null as locale in the stub intl classes
The Intl extension accepts null as locale in formatters and the collator and will use the default locale in such case. Given that the stub implementation considers that the default locale is always 'en', it should be supported here too instead of forcing libraries to pass the default locale explicitly. For instance, ``Twig_Extensions_Extension_Intl`` relies on the fact that ``null`` as locale uses the default one: https://github.com/twigphp/Twig-extensions/pull/94
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | https://github.com/symfony/symfony-demo/pull/123
| License | MIT
| Doc PR | n/a
Commits
-------
d6db6ad Fix the handling of null as locale in the stub intl classes
Many tests being skipped based on the ICU data version don't actually
need it. They might be testing code paths not relying on Intl, or not
performing assertions on the values depending on the ICU data and so not
dependant on the exact ICU version being used.
The Intl extension accepts null as locale in formatters and the
collator and will use the default locale in such case. Given that the
stub implementation considers that the default locale is always 'en', it
should be supported here too instead of forcing libraries to pass the
default locale explicitly.
This PR was squashed before being merged into the 2.3 branch (closes#15172).
Discussion
----------
[DependencyInjection] fixed FrozenParameterBag and improved Parameter…
The ParameterBagInterface was missing some @throws annotations, so the FrozenParameterBag class was a violation of Liskov subtitution principle. Also the ParameterBagInterface was missing the remove method.
(Optionally the ParameterBagInterface can be later split into two smaller interfaces, because the FrozenParameterBag shouldn't have the add, remove methods in the first place.)
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | yes
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
I have also fixed removing elements from FrozenParameterBag, as introduced by @satahippy
https://github.com/symfony/DependencyInjection/pull/8
Commits
-------
3ad0794 [DependencyInjection] fixed FrozenParameterBag and improved Parameter…
This PR was squashed before being merged into the 2.3 branch (closes#15249).
Discussion
----------
[HttpFoundation] [PSR-7] Allow to use resources as content body and to return resources from string content
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | not yet
* Allows to fix tests of https://github.com/symfony/psr-http-message-bridge with PHP 5.6.
* Ease the transition to PSR-7 (in PSR-7, almost everything is stream - #15186)
Maybe should I open it against 2.8 but it can be considered a bug fix at least for the part "returning a string as a resource".
Commits
-------
059964d [HttpFoundation] [PSR-7] Allow to use resources as content body and to return resources from string content
This PR was merged into the 2.3 branch.
Discussion
----------
[HttpFoundation] Behaviour change in PHP7 for substr
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
In PHP7 the behaviour of substr() changed.
To resume: "Truncating an entire string should result in a string."
See: https://bugs.php.net/bug.php?id=62922
Commits
-------
ba6000b [HttpFoundation] Behaviour change in PHP7 for substr
In PHP7 the behaviour of substr() changed.
To resume: "Truncating an entire string should result in a string."
See: https://bugs.php.net/bug.php?id=62922
This PR was squashed before being merged into the 2.3 branch (closes#15223).
Discussion
----------
[Finder] Command::addAtIndex() fails with Command instance argument
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #14384
| License | MIT
| Doc PR | -
Fixed reported bug #14384 and added test case for it. This is a second PR as previous #14385 went bad after I failed to rebase 2.3 branch properly.
Commits
-------
2aff566 [Finder] Command::addAtIndex() fails with Command instance argument
This PR was merged into the 2.3 branch.
Discussion
----------
CS fixes for 2.3
| 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
To keep fabbot.io happy ;)
Commits
-------
464b67e fix CS
This PR was merged into the 2.3 branch.
Discussion
----------
Mock microtime() and time() in transient tests
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Commits
-------
8319ca3 Mock microtime() and time() in transient tests
In AZ, as in TR, pluralization is always 0:
0 kitab (zero books)
1 kitab (1 book)
3 kitab (3 books)
104 kitab (104 books)
Apparently ZF ruleset was wrong in the first place :)
This PR was merged into the 2.3 branch.
Discussion
----------
[Form] Fixed: Data mappers always receive forms indexed by their names
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
This PR facilitates writing domain-specific data mappers, since it guarantees that you can access forms by name in the data mapper methods. Currently, `Form::add()` does not set the index of the array passed to the data mapper to the form's name.
Commits
-------
86b7fe5 [Form] Fixed: Data mappers always receive forms indexed by their names
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 merged into the 2.3 branch.
Discussion
----------
[DependencyInjection] Fail when dumping a Definition with no class nor factory
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
I tried creating an anonymous inline service with a DefinitionDecorator, but that did not work.
Here is the fix.
Commits
-------
23ad4ad [DependencyInjection] Fail when dumping a Definition with no class nor factory
This PR was merged into the 2.3 branch.
Discussion
----------
[Form] fixed sending non array data on submit to ResizeListener
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #13804, #13851
| License | MIT
| Doc PR |
I agree with @Tobion that #13851 is a bug fix and therefore should be applied on the `2.3` branch too.
Commits
-------
0f185c9 [Form] [EventListener] fixed sending non array data on submit to ResizeListener
This PR was merged into the 2.3 branch.
Discussion
----------
[HttpFoundation] Use convention to allow throwing from __toString()
| Q | A
| ------------- | ---
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Corollary to #15076, works without it.
Commits
-------
8982c32 [HttpFoundation] Use convention to allow throwing from __toString()
This PR was squashed before being merged into the 2.3 branch (closes#15058).
Discussion
----------
[Console] Fix STDERR output text on IBM iSeries OS400
| Q | A
| ------------- | ---
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | None
| License | MIT
| Doc PR | None
Prior to this PR a Symfony Console command would output error text as random symbols when executed via the IBM console program QSH. Affected error text output such as when a required InputArgument is missing, or when explicitly using `$output->getErrorOutput()->writeln('Some error text here')`.
![Example error output](http://i.imgur.com/PQplK1p.png)
This PR fixes error text so it properly prints to IBM console programs such as QSH and QP2SHELL.
I previously fixed STDOUT for PHP running on IBM iSeries OS400 (Zend Server) using the same approach. Since that PR was merged ConsoleOutput class began using its own output for STDERR which exhibits the same issue STDOUT did.
The following commits and previous Symfony PRs have our relevant discussion about ASCII vs EBCDIC character encoding to fix this issue:
* [Original IBM STDOUT reported in #1434](https://github.com/symfony/symfony/issues/1434)
* [My PR #4152 that fixes#1434](https://github.com/symfony/symfony/pull/4152)
Thanks!
🚀
Commits
-------
23c42ca [Console] Fix STDERR output text on IBM iSeries OS400