* 2.7:
[TwigBundle] fixed usage when Templating is not installed
[Validator] Check cascasdedGroups for being countable
[Filesystem] Check that the directory is writable after created it in dumpFile()
[HttpFoundation] Improved set cookie header tests
[Console] increased code coverage of Output classes
This PR was squashed before being merged into the 2.7 branch (closes#21155).
Discussion
----------
[Validator] Check cascasdedGroups for being countable
Prevents notice for PHP 7.2
| 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 |
Just repeated for this place https://github.com/symfony/symfony/pull/20859
Waiting travis to fill in the `Tests pass` field
`cascasdedGroups` can be null at this point (according to failures and phpdoc)
Commits
-------
8fa45a130b [Validator] Check cascasdedGroups for being countable
This PR was merged into the 3.2 branch.
Discussion
----------
[Bridge/PhpUnit] Relax expectedDeprecation for forward compat
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Instead of a strict match, this makes the annotation check ignore not expected messages, while still requiring all the expected ones in order.
This is needed for forward compat with future deprecations, and should make travis green again.
Commits
-------
a3ba726 [Bridge/PhpUnit] Relax expectedDeprecation for forward compat
This PR was merged into the 3.2 branch.
Discussion
----------
[FrameworkBundle] Remove Response* from classes to compile
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20560, https://github.com/symfony/symfony-demo/issues/411, https://github.com/FriendsOfSymfony/FOSHttpCacheBundle/issues/276
| License | MIT
| Doc PR | -
When HttpCache is used, Response is loaded first, then the kernel is booted (on cache miss), which triggers the loading of classes.php. Since 3.2 generates a context free classes.php, the Response class is now included there when it was excluded previously. And boom, "Cannot declare class Symfony\Component\HttpFoundation\Response".
Commits
-------
9ab59824a1 [FrameworkBundle] Remove Response* from classes to compile
This PR was squashed before being merged into the 2.7 branch (closes#20972).
Discussion
----------
[HttpFoundation] Improved set cookie header tests
| Q | A
| ------------- | ---
| Branch? | 2.7 (already in master)
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #... <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | symfony/symfony-docs#... <!--highly recommended for new features-->
Separated from #20569
Commits
-------
05bce71d7a [HttpFoundation] Improved set cookie header tests
This PR was squashed before being merged into the 3.1 branch (closes#21165).
Discussion
----------
[Serializer] int is valid when float is expected when deserializing JSON
| Q | A
| ------------- | ---
| Branch? | 3.1
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | api-platform/api-platform#37
| License | MIT
| Doc PR | n/a
JSON only has a Number type corresponding to both `int` and `float` PHP types.
PHP's `json_encode`, JavaScript's `JSON.stringify`, Go's `json.Marshal` as well as most other JSON encoders convert floating-point numbers like `12.0` to `12` (the decimal part is dropped when possible).
PHP's `json_decode` automatically converts Numbers without a decimal part to integers.
Actually, the Serializer rejects integers when a float is expected, this PR fixes this behavior when denormalizing JSON-based formats.
Port of api-platform/core#714.
/cc @gorghoa @Shine-neko
Commits
-------
4125455775 [Serializer] int is valid when float is expected when deserializing JSON
This PR was squashed before being merged into the 2.7 branch (closes#21124).
Discussion
----------
[Console] increased code coverage of Output classes
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
This PR increases the coverage of Output classes of the Console component from 80.81% to 94.95%
Commits
-------
ab4ba23 [Console] increased code coverage of Output classes
This PR was merged into the 2.8 branch.
Discussion
----------
[Profiler][VarDumper] Fix minor color issue & duplicated selector
| Q | A
| ------------- | ---
| Branch? | 2.8
| 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 working on #21109, I spotted this minor issue with `sf-dump-const` and `.sf-dump-ref`:
| Before | After |
| --- | --- |
|<img width="276" alt="screenshot 2017-01-03 a 20 13 55" src="https://cloud.githubusercontent.com/assets/2211145/21619779/7e1e347e-d1f1-11e6-9d84-fbb1d5d6b1fa.PNG">| <img width="275" alt="screenshot 2017-01-03 a 20 14 04" src="https://cloud.githubusercontent.com/assets/2211145/21619786/86dde5dc-d1f1-11e6-8b13-dcfc5abe466a.PNG">|
Commits
-------
b282076 [Profiler][VarDumper] Fix minor color issue & duplicated selector
This PR was merged into the 3.2 branch.
Discussion
----------
[Cache] Remove silenced warning tiggered by PhpArrayAdapter
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Because many reported this silenced warning, which happens all the time unless `cache:warmup` has been run.
Commits
-------
a9dfcd8 [Cache] Remove silenced warning tiggered by PhpArrayAdapter
This PR was merged into the 3.2 branch.
Discussion
----------
[Console] Fix return type in Terminal docblocks
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
This PR updates the docblock to reflect what's already in the code.
Commits
-------
6afb0afa24 [Console] Fix return type in Terminal docblocks
This PR was merged into the 3.2 branch.
Discussion
----------
[DependencyInjection] Fix the priority order of compiler pass trait
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20332, #20993
| License | MIT
| Doc PR | ~
A regression has appeared between the version `3.1` and `3.2` on the compilation passes using a priority option (see #20332).
Indeed, the order is no longer preserved for service definitions with the same priority. This is caused by the `SplPriorityQueue` class that does not have a FIFO implementation (see this [comment](https://github.com/symfony/symfony/issues/20332#issuecomment-268096527)).
The PR #20993 fixes the problem but only for Forms, not for all compiler passes using the `PriorityTaggedServiceTrait` trait since `3.2`.
Commits
-------
aac9a7e Fix the priority order of compiler pass trait
This PR was merged into the 3.2 branch.
Discussion
----------
[PhpUnitBridge] Set COMPOSER_ROOT_VERSION while installing
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Commits
-------
8bab09e [PhpUnitBridge] Set COMPOSER_ROOT_VERSION while installing
This PR was merged into the 2.7 branch.
Discussion
----------
[HttpFoundation] Validate/cast cookie expire time
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #... <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | symfony/symfony-docs#... <!--highly recommended for new features-->
Commits
-------
8215dbdb31 [HttpFoundation] Validate/cast cookie expire time
This PR was merged into the 3.2 branch.
Discussion
----------
[PhpUnitBridge] skip tests with failure and error states too
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
I was actually wrong by not excluding failures and errors to in #21121.
Commits
-------
f3bb206f12 skip tests with failure and error states too
This PR was merged into the 3.2 branch.
Discussion
----------
[Yaml] add missing indicator character
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
The comma was mentioned in the deprecation message, though it actually was never checked before in the condition.
Commits
-------
8725e0b [Yaml] add missing indicator character
* 3.1:
do not depend on a fixed date in layout tests
[Console] Escape default value when dumping help
[Console] OS X Can't call cli_set_process_title php without superuser
Fixed @return when returning this or static #bis
fixed @return when returning this or static
Polish translation improvement in Validator component
[Console] Descriptors should use Helper::strlen
[Config] Improve PHPdoc / IDE autocomplete
[Debug] Wrap call to ->log in a try catch block
[Debug] UndefinedMethodFatalErrorHandler - Handle anonymous classes
fix merge
[cache] Bump RedisAdapter timeout to 5s
fixed @return when returning this or static
[SecurityBundle] Made collection of user provider unique when injecting them to the RemberMeService
remove is_writable check on filesystem cache
* 2.8:
do not depend on a fixed date in layout tests
[Console] Escape default value when dumping help
[Console] OS X Can't call cli_set_process_title php without superuser
Fixed @return when returning this or static #bis
Polish translation improvement in Validator component
[Console] Descriptors should use Helper::strlen
[Config] Improve PHPdoc / IDE autocomplete
[Debug] Wrap call to ->log in a try catch block
[Debug] UndefinedMethodFatalErrorHandler - Handle anonymous classes
[SecurityBundle] Made collection of user provider unique when injecting them to the RemberMeService
* 2.7:
do not depend on a fixed date in layout tests
[Console] Escape default value when dumping help
[Console] OS X Can't call cli_set_process_title php without superuser
Polish translation improvement in Validator component
[Console] Descriptors should use Helper::strlen
[Config] Improve PHPdoc / IDE autocomplete
[Debug] Wrap call to ->log in a try catch block
[Debug] UndefinedMethodFatalErrorHandler - Handle anonymous classes
[SecurityBundle] Made collection of user provider unique when injecting them to the RemberMeService
This PR was merged into the 3.2 branch.
Discussion
----------
[DI] Minor fixes in FactoryReturnTypePass
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Commits
-------
c231f81 [DI] Minor fixes in FactoryReturnTypePass
By default, the `DateType` as well as the `DateTimeType` set the choices
being available for the year to a range starting five years in the past.
After some time, this will make tests fail when the year of the fixed
date being used as the initial data is before the first year being part
of the choices.
This PR was merged into the 2.7 branch.
Discussion
----------
[SecurityBundle] Made collection of user provider unique when injecting them to the RemberMeService
| 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 | -
Commits
-------
8227593 [SecurityBundle] Made collection of user provider unique when injecting them to the RemberMeService
This PR was merged into the 2.7 branch.
Discussion
----------
[Console] OS X Can't call cli_set_process_title php without superuser
| 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
The console component test suite is failing on OS X (10.12.2) as the `cli_set_process_title` function seems not to be allowed to be called without superuser. It seems to be an OS X limitation.
At least, the test suite should pass and a warning should be shown when using the component with this feature on OS X.
---
refs:
- http://stackoverflow.com/questions/27803120/cannot-set-process-title-in-a-php-command-line-script-using-cli-set-process-titl
- https://github.com/liip/php-osx/issues/139
- https://github.com/zfcampus/zf-console/pull/22
Commits
-------
9928c0d [Console] OS X Can't call cli_set_process_title php without superuser
This PR was merged into the 3.2 branch.
Discussion
----------
[TwigBridge] Fix upgrade/changelog notes
| Q | A
| ------------- | ---
| Branch? | 3.2
The current upgrade note is wrong, the renderer cannot be injected in the `TwigRendererEngine`, only the Twig Environment can be.
Also added a missing note about the deprecated `TwigRendererEngineInterface`.
Commits
-------
a7ebe9c [TwigBridge] Fix upgrade/changelog notes
This PR was squashed before being merged into the 2.7 branch (closes#20900).
Discussion
----------
[Console] Descriptors should use Helper::strlen
| 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
-------
8e9a5f8 [Console] Descriptors should use Helper::strlen
This PR was merged into the 2.7 branch.
Discussion
----------
[Config] Improve PHPdoc / IDE autocomplete
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes-ish
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #... <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | symfony/symfony-docs#... <!--highly recommended for new features-->
The missing pieces for fixing the autocomplete in IDE's (tested in `PhpStorm 2016.3.2`).
Each method seems to be found now for the framework bundle configuration 😱
Actually uses #20923 where needed, but i think we should go with it consistently.
Relates to https://github.com/symfony/symfony/pull/20290#issuecomment-257303145
Commits
-------
3f3c6e0 [Config] Improve PHPdoc / IDE autocomplete
This PR was squashed before being merged into the 2.7 branch (closes#21010).
Discussion
----------
[Debug] UndefinedMethodFatalErrorHandler - Handle anonymous classes
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
When trying to call a method on an anonymous class (with or without methods) the `UndefinedMethodFatalErrorHandler` tries to iterate `null`.
For ref.: https://3v4l.org/l26u1
Commits
-------
ed713ae [Debug] UndefinedMethodFatalErrorHandler - Handle anonymous classes
This PR was merged into the 3.1 branch.
Discussion
----------
[cache] Bump RedisAdapter default timeout to 5s
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
Bump RedisAdapter timeout to 5s because (at least with Predis) 0 means 0s
Commits
-------
eaca2a4 [cache] Bump RedisAdapter timeout to 5s
* 3.1:
fixed @return when returning this or static
override property constraints in child class
removed unneeded comment
[Console] improved code coverage of Command class
[FrameworkBundle] Make TemplateController working without the Templating component
Only count on arrays or countables to avoid warnings in PHP 7.2
* 2.8:
fixed @return when returning this or static
override property constraints in child class
removed unneeded comment
[Console] improved code coverage of Command class
[FrameworkBundle] Make TemplateController working without the Templating component
Only count on arrays or countables to avoid warnings in PHP 7.2
* 2.7:
fixed @return when returning this or static
override property constraints in child class
[Console] improved code coverage of Command class
Only count on arrays or countables to avoid warnings in PHP 7.2
This PR was merged into the 2.7 branch.
Discussion
----------
Fix @return statements to use $this or static when relevant
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20290
| License | MIT
| Doc PR | n/a
see #20290
Commits
-------
3c0693d fixed @return when returning this or static
This PR was squashed before being merged into the 3.2 branch (closes#20959).
Discussion
----------
[FrameworkBundle] Ignore AnnotationException exceptions in the AnnotationsCacheWarmer
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
I ran into a case that led to an `AnnotationException` in the `AnnotationsCacheWarmer`. I deduced that these exceptions should be ignored to not break the cache warming process.
The case :
* You have a controller in your `AppBundle\Controller\` directory.
* You have an annotation in this controller that will never be found (because you made a mistake in the class / annotation name) or not every time : for example your controller is only used in dev env and it uses an annotation that is defined in a vendor bundle that is only under the `require-dev` key in your `composer.json` because you don't need it in your production deployment. So in production, your controller exists but will never be used. However the vendor bundle does not exist.
* You dump your autoload files with composer using the `--optimize` option.
* You end up with an `AnnotationException`.
Why :
* In the [`FrameworkExtension`](https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/DependencyInjection/FrameworkExtension.php#L189), the `**Bundle\Controller\` pattern is added as an "annotated class to compile".
* In the [`AddClassesToCachePass`](https://github.com/symfony/symfony/blob/master/src/Symfony/Component/HttpKernel/DependencyInjection/AddClassesToCachePass.php#L50), all the classes defined in the `autoload_classmap.php` file are matched against the previous pattern.
* Your controller will match. So even if you don't use it at all in your application, the annotation will still be read and the exception will be thrown.
Commits
-------
6749569 [FrameworkBundle] Ignore AnnotationException exceptions in the AnnotationsCacheWarmer
This PR was merged into the 3.2 branch.
Discussion
----------
[FrameworkBundle] Allow multiple transitions with the same name
| Q | A
| ------------- | ---
| Branch? | 3.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20794
| License | MIT
| Doc PR | -
Commits
-------
7c86e16 [FrameworkBundle] Allow multiple transactions with the same name
This PR was merged into the 2.7 branch.
Discussion
----------
Avoid warning in PHP 7.2 because of non-countable data
| 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 | -
Recently, the "[Counting of non-countable objects][1]" RFC was merged in PHP 7.2-dev. This means `count()` now causes a warning when passing anything that's not countable (e.g. `null` or `''`).
As PHP does not lazily execute conditions, `FormUtil::isEmtpy($data) || 0 === count($data)` will cause *both* conditions to be executed. This means `count($data)` errors when `$data` is empty.
Splitting it up in 2 statements avoids the warning being triggered in PHP 7.2.
See https://travis-ci.org/symfony-cmf/content-bundle/jobs/181815895 for a failing test caused by this bug.
[1]: https://wiki.php.net/rfc/counting_non_countables
Commits
-------
94253e8 Only count on arrays or countables to avoid warnings in PHP 7.2
This PR was squashed before being merged into the 2.7 branch (closes#20817).
Discussion
----------
[Console] improved code coverage of Command class
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
This PR increases the code coverage of the Command class from ``86.82%`` to ``96.90%``.
Commits
-------
d393113 [Console] improved code coverage of Command class
This PR was merged into the 2.8 branch.
Discussion
----------
[FrameworkBundle] Make TemplateController working without the Templating component
| Q | A
| ------------- | ---
| Branch? | 2.8
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
The template controller was missed during #15502. Btw I've added some tests for this controller.
Commits
-------
89ccbc4 [FrameworkBundle] Make TemplateController working without the Templating component
* 3.1:
use HHVM 3.15 to run tests
[TwigBridge] fix Twig 2.x compatibility
removed some PHP CS Fixer rules
Improve language
[Console] SymfonyStyle: Escape trailing backslashes in user texts
Mention the community review guide
[Form] fix group sequence based validation
[Console] Fix question formatting using SymfonyStyle::ask()
* 2.8:
use HHVM 3.15 to run tests
[TwigBridge] fix Twig 2.x compatibility
removed some PHP CS Fixer rules
Improve language
[Console] SymfonyStyle: Escape trailing backslashes in user texts
Mention the community review guide
[Form] fix group sequence based validation
[Console] Fix question formatting using SymfonyStyle::ask()
* 2.7:
use HHVM 3.15 to run tests
[TwigBridge] fix Twig 2.x compatibility
removed some PHP CS Fixer rules
Improve language
[Console] SymfonyStyle: Escape trailing backslashes in user texts
Mention the community review guide
[Form] fix group sequence based validation
[Console] Fix question formatting using SymfonyStyle::ask()
This PR was merged into the 2.7 branch.
Discussion
----------
[Console] Fix question formatting using SymfonyStyle::ask()
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20964
| License | MIT
| Doc PR | n/a
Given
```php
$io = new SymfonyStyle($input, $output);
$io->ask('Do you want to use Foo\\Bar <comment>or</comment> Foo\\Baz\\?', 'Foo\\Bar');
```
Before output
![before](http://image.prntscr.com/image/af3806f866654deda2dec79b50d1ffa2.png)
After output
![after](http://image.prntscr.com/image/59c031d9e02949cebeae7a4734c7043f.png)
Commits
-------
e189183 [Console] SymfonyStyle: Escape trailing backslashes in user texts
9d46712 [Console] Fix question formatting using SymfonyStyle::ask()
This PR was merged into the 2.7 branch.
Discussion
----------
[Form] fix group sequence based validation
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20929
| License | MIT
| Doc PR |
Commits
-------
fb91f74 [Form] fix group sequence based validation
* 3.1:
fixed obsolete getMock() usage
fixed obsolete getMock() usage
fixed obsolete getMock() usage
[WebProfilerBundle] Display multiple HTTP headers in WDT
do not remove the Twig ExceptionController service
removed obsolete condition
do not try to register incomplete definitions
* 2.8:
fixed obsolete getMock() usage
fixed obsolete getMock() usage
[WebProfilerBundle] Display multiple HTTP headers in WDT
do not remove the Twig ExceptionController service
removed obsolete condition
do not try to register incomplete definitions