This PR was squashed before being merged into the master branch (closes#6491).
Commits
-------
5d264ce [Routing] made RouteCompilerInterface::compile static
Discussion
----------
[Routing] made RouteCompilerInterface::compile static
bc break: yes but probably nobody affected
It makes much more sense that the compile method is static because having instances of a compiler is strange. Also the compiler could not use DI or anything anyway as the `Route` class was instantiating compiler instances itself and was assuming there are no constructor arguments.
This PR was squashed before being merged into the master branch (closes#6594).
Commits
-------
9d94fc7 [Console] Autocomplete cleanups/optimisations
Discussion
----------
[Console] Autocomplete cleanups/optimisations
I know the optimisations to this code are getting kinda ridiculous, but I found a few more things to tidy up.
This PR was merged into the master branch.
Commits
-------
f5290b9 [FrameworkBundle] Force users to set "kernel.secret" to something different than default "ThisTokenIsNotSoSecretChangeIt"
Discussion
----------
[RFC][BC][FrameworkBundle] Force users to set "kernel.secret" to something unique
Bug fix: kinda*
Feature addition: no
BC break: yes
Symfony2 tests pass: yes
Fixes the following tickets: #6480
License of the code: MIT
This PR is to show different approach for "fix" suggested in #6480, as IMO there is no real point for "yet another listener" =)
This PR also introduces BC break for all users that used default value for `kernel.secret`, but IMO it's worth it.
This PR was merged into the master branch.
Commits
-------
4bf82e0 Update Ukrainian translation for master branch
Discussion
----------
[Validator] Update Ukrainian translation for master branch
This PR was merged into the master branch.
Commits
-------
2b73975 Add new tests and fix problem with the second option being chosen on down arrow
8a0bcfb Use strpos in place of substr
9864d95 Not much use in blanking the array each time
a4d711c Enable arrow keys to browse matched autocomplete options
Discussion
----------
[Console] Enable arrow keys to browse matched autocomplete options
See notes in original autocomplete pull request: #6391
See also @bamarni's pull request implementing more or less the same stuff: #6561
---------------------------------------------------------------------------
by fabpot at 2013-01-05T10:12:47Z
Looks good to me. Can you add some unit tests?
---------------------------------------------------------------------------
by bamarni at 2013-01-05T12:51:51Z
1 line addition... you definitely got me on the diff!
That's what I had mind too, excepted for the default highlight, but as you said it's usually displayed in the question so it 's not necessary. 👍
---------------------------------------------------------------------------
by lmcd at 2013-01-06T04:07:02Z
@fabpot Added tests
* 2.1:
[Console] made Application::getTerminalDimensions() public
Revert "merged branch egeloen/f-2.0-terminal-width (PR #6571)"
[2.1] [Console] Added getTerminalDimensions() with fix for osx/freebsd
Restrict Monolog version to be in version `<1.3`. Because of conflict between `HttpKernel/Log/LoggerInterface` and `Psr\Log\LoggerInterface` (PSR-3)
Conflicts:
composer.json
This PR was merged into the master branch.
Commits
-------
e0b4480 [Form] Removed separator characters between choice or text fields in DateType
Discussion
----------
[Form] Removed separator characters between choice or text fields in DateType
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -
License of the code: MIT
Documentation PR: -
This commit was originally contained in #6217 (for 2.1) but then split away for master since it changes behavior.
This PR was merged into the master branch.
Commits
-------
184c8e5 Fixed @expectedException definitions to reference absolute exception paths
Discussion
----------
Added leading slashes to @expectedException definitions
Bug fix: no
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -
License of the code: MIT
Documentation PR: -
This PR adds leading backslashes `\` to `@expectedException` declarations.
The current version is inconsistent with other class references in doc blocks, which default to *relative* class names unless class name has a leading backslash.
```php
/**
* @param RelativeNs\Class $param1
* @param \AbsoluteNs\Class $param2
*/
but
/**
* @expectedException AbsoluteNs\Class
*/
```
Consequently, PHPStorm does not understand the current statements and marks them as erroneous (and correctly so IMO, even though PHPUnit's interpretation is more relaxed).
* 2.1:
Restrict Monolog version to be in version <1.3
[Console] Make getTerminalWith & getTerminalHeight public
[DependencyInjection] fixed PhpDumper optimizations when an inlined service depends on the current one indirectly
[DependencyInjection] fixed PhpDumper when an inlined service definition has some properties
[DependencyInjection] added some tests for PhpDumper when the container is compiled
[DependencyInjection] fixed CS
[Process] Do not reset stdout/stderr pipes on Interrupted system call
[Locale] Adjust `StubIntlDateFormatter` to have new methods added in PHP 5.5
use the right RequestMatcherInterface
[Locale] Fix failing `StubIntlDateFormatter` tests in PHP 5.5
[Locale] Fix failing `StubIntlDateFormatter` in PHP 5.5
[Form] Fix failing `MonthChoiceList` in PHP 5.5
Update .travis.yml
Conflicts:
src/Symfony/Bridge/Monolog/composer.json
src/Symfony/Component/DependencyInjection/Tests/Fixtures/php/services9.php
* 2.0:
Restrict Monolog version to be in version <1.3
[Console] Make getTerminalWith & getTerminalHeight public
[DependencyInjection] fixed PhpDumper optimizations when an inlined service depends on the current one indirectly
[DependencyInjection] fixed PhpDumper when an inlined service definition has some properties
[DependencyInjection] added some tests for PhpDumper when the container is compiled
[DependencyInjection] fixed CS
[Locale] Adjust `StubIntlDateFormatter` to have new methods added in PHP 5.5
[Locale] Fix failing `StubIntlDateFormatter` tests in PHP 5.5
[Locale] Fix failing `StubIntlDateFormatter` in PHP 5.5
[Form] Fix failing `MonthChoiceList` in PHP 5.5
Update .travis.yml
Conflicts:
src/Symfony/Bridge/Monolog/composer.json
src/Symfony/Component/DependencyInjection/Tests/Fixtures/php/services9.php
src/Symfony/Component/Form/Extension/Core/ChoiceList/MonthChoiceList.php
tests/Symfony/Tests/Component/DependencyInjection/Fixtures/yaml/services9.yml
This PR was merged into the 2.0 branch.
Commits
-------
e0923ae [DependencyInjection] fixed PhpDumper optimizations when an inlined service depends on the current one indirectly
cd15390 [DependencyInjection] fixed PhpDumper when an inlined service definition has some properties
e939a42 [DependencyInjection] added some tests for PhpDumper when the container is compiled
3827e3e [DependencyInjection] fixed CS
Discussion
----------
[DependencyInjection] Fixed PhpDumper when compiling inlined services
This PR was merged into the master branch.
Commits
-------
97bc257 [Locale] Rolled back normalizeIcuVersion() method in unit tests.
Discussion
----------
[Locale] Rolled back normalizeIcuVersion() method in unit tests.
We just need to check if version is 4.4 or 4.8 and we do not need to know if it is 4.8.0 or 4.8.1.1 (at least for now there are no big differences between minor ICU versions).
This change basically rolls back modifications made in #6109 to fix#5288. Those modifications didn't really fix the issue since ICU>=4.9 wasn't handled properly. Part of the roll back was made in #6515 (2.1 branch).
This change makes master branch consistent with what we have in 2.0 and 2.1 (bespoke tests are passing in those branches).
Bug fix: no
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: one test fails in the Propel1 bridge
Fixes the following tickets: #5517
Todo: -
License of the code: MIT
Documentation PR: -
This PR was merged into the 2.0 branch.
Commits
-------
f299bd0 [Console] Make getTerminalWith & getTerminalHeight public
Discussion
----------
[Console] Make getTerminalWith & getTerminalHeight public
Bug fix: yes
Feature addition: yes
Backwards compatibility break: no
Fixes the following tickets: ~
Todo: -
License of the code: MIT
Documentation PR: ~
After demand in #6567, I open an other one on the 2.0 branch.
This makes it consistent with 2.0 and 2.1.
We just need to check if version is 4.4 or 4.8 and we do not need to know if it is 4.8.0 or 4.8.1.1 (at least for now there are no big differences between minor ICU versions).
This PR was merged into the 2.1 branch.
Commits
-------
fa6fb6f [Process] Do not reset stdout/stderr pipes on Interrupted system call
Discussion
----------
[Process][2.1] Do not reset stdout/stderr pipes on stream_select failure
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
License of the code: MIT
When `stream_select` is called and failed, pipes are currently reset to an empty array. As a result, both error-output and output are empty whereas the exitcode is 0 (successful process).
There is no reason (at least I don't see it) to reset pipes when an error occurs.
---------------------------------------------------------------------------
by fabpot at 2013-01-04T08:18:03Z
This behavior was introduced by @Seldaek in bcf8cf91. Any reason?
---------------------------------------------------------------------------
by Seldaek at 2013-01-04T17:22:33Z
The rationale was that if stream_select failed, it will fail every time (maybe this is an incorrect assumption and I introduced a bug there). Since the while loop checks for $this->pipes being not empty, if the process dies off and stream_select() starts failing, it will be stuck in the while loop forever. That's what I was fixing there. As far as I can tell, your fix reintroduces that problem. Can you explain a bit more what you mean by "error-output and output are empty"? Maybe adding a test for the case you are seeing would help too?
---------------------------------------------------------------------------
by romainneutron at 2013-01-04T17:47:00Z
Hi !
I'm actually using `Process` in a amqp client using PhpAmqpLib. Sometimes both output and error-output of a process are empty whereas the process returned with an exitcode equal to 0.
By *empty*, I mean that the output of both `$process->getOuptut()` and `$process->getErrorOuptut()` are equal to `''`.
I removed the `@` before the `stream_select` and realized that some calls where failing :
```
stream_select(): unable to select [4]: Interrupted system call (max_fd=13) in .../vendor/symfony/process/Symfony/Component/Process/Process.php on line 341
```
With my patch, it appears that if a `stream_select` call fails one time, it can be called with success a second time.
Adding a test case that triggers this error does not seem easy as it seems system related.
What do you think about it ?
---------------------------------------------------------------------------
by Seldaek at 2013-01-04T17:52:04Z
Ok I see. So I guess your fix makes sense since stream_select can fail legitimately, but it's not the whole story. If it fails because the process died off (or whatever the cause was on windows) we shouldn't keep trying forever. Maybe just adding a limit of N retries that is reset whenever a call succeeds would be enough to go over the odd system call being interrupted without allowing infinite loops. Not sure what N makes sense.
---------------------------------------------------------------------------
by igorw at 2013-01-04T17:53:45Z
As per the `stream_select` manpage it can return false when interrupted by a system call. That does not indicate that something is broken. However, I'm not sure if it is possible to distinguish between actual errors and interrupts without capturing and parsing the error message.
---------------------------------------------------------------------------
by romainneutron at 2013-01-04T18:50:56Z
I've updated the PR to consider the error message as it's currently the only solution to find out why `stream_select` failed
---------------------------------------------------------------------------
by romainneutron at 2013-01-04T18:59:15Z
Merged fixes suggested by @stloyd and @Seldaek
When `stream_select` is called and return false because of an `Interrupted
system call`, pipes are currently reset to an empty array.
As a result, both error-output and output are empty whereas the
exitcode is 0 (successful process).
This PR was merged into the 2.0 branch.
Commits
-------
73d9cef [Locale] Adjust `StubIntlDateFormatter` to have new methods added in PHP 5.5
b2ce983 [Locale] Fix failing `StubIntlDateFormatter` tests in PHP 5.5
913b564 [Locale] Fix failing `StubIntlDateFormatter` in PHP 5.5
8ae773b [Form] Fix failing `MonthChoiceList` in PHP 5.5
f4ce2f1 Update .travis.yml
Discussion
----------
[2.0] Add testing of PHP 5.5 on Travis-CI & fix failing code
