Doctrine caches annotations. For methods, it uses PHP reflection and the getDeclaringClass() to create
a unique cache key. Unfortunately, if you have 2 classes that extend another one, the cache will be shared.
It's not a problem except that before this patch, the default route name was also cached (as the cache is serialized
after we changed the object). So, all other classes inherited this default route name. The fix is quite easy:
just don't change the read annotation object.
From the PHP CHANGELOG:
The flag ENT_SUBSTITUTE makes invalid multibyte sequences be replaced by
U+FFFD (UTF-8) or &#FFFD; by htmlspecialchars and htmlentities. It is an
alternative to the default behavior, which just returns an empty string and to
ENT_IGNORE, which is a security risk. The behavior follows the recommendations
of Unicode Technical Report #36.
Commits
-------
ae3b128 [ClassLoader] Support for autoloading include_path incl. tests.
Discussion
----------
Autoload
GH Issue #1823
---------------------------------------------------------------------------
by stof at 2011/07/29 00:42:10 -0700
note that another fix was proposed in #1852 but this implementation is cleaner IMO
---------------------------------------------------------------------------
by henrikbjorn at 2011/08/12 01:57:45 -0700
@fabpot @stof any suggestions? need this kind of badly
---------------------------------------------------------------------------
by stof at 2011/08/12 02:06:54 -0700
for me it is fine. I guess you need to wait the end of @fabpot's holydays to see it merged.
---------------------------------------------------------------------------
by henrikbjorn at 2011/08/25 02:24:13 -0700
Added tests in the hope it will make it in soon :)
---------------------------------------------------------------------------
by henrikbjorn at 2011/08/29 03:31:08 -0700
Any other requests / suggestions ?
---------------------------------------------------------------------------
by stof at 2011/08/29 03:36:15 -0700
could you rebase the PR ? Github says that it conflicts.
---------------------------------------------------------------------------
by henrikbjorn at 2011/08/29 04:11:43 -0700
Should be rebased now or that is what git cli says :)
---------------------------------------------------------------------------
by henrikbjorn at 2011/08/29 04:16:28 -0700
And squashed.
Commits
-------
6278fcb -- add dumpers for translation component
Discussion
----------
[2.1] Add dumpers for translation catalogs
As seen here #1283, I push just the translation catalogs dumpers. It involved renaming and some essentially.
I also included a Pot dumper/loader that have been provided here : https://github.com/michelsalib/BCCExtraToolsBundle/pull/12.
---------------------------------------------------------------------------
by fabpot at 2011/08/28 11:12:30 -0700
Can you add the license header? and the main phpdoc block for each class?
---------------------------------------------------------------------------
by michelsalib at 2011/08/29 02:32:50 -0700
Done !
---------------------------------------------------------------------------
by fabpot at 2011/08/29 03:17:43 -0700
Last, but not the least, can you add some unit tests and squash all your commits? Thanks a lot.
---------------------------------------------------------------------------
by michelsalib at 2011/08/29 03:21:43 -0700
How do I squash it, should I make a new PR ?
---------------------------------------------------------------------------
by fabpot at 2011/08/29 03:25:09 -0700
No need to make a new PR, you can just force the push with `-f`.
---------------------------------------------------------------------------
by fabpot at 2011/08/29 03:33:59 -0700
Also, in the tests, it would great if you can add some that do a load/dump/load, just to be sure that everything work fine.
---------------------------------------------------------------------------
by michelsalib at 2011/08/29 03:39:41 -0700
What is the need of such a test, considering that the current tests that I am writing are using the same fixtures ?
---------------------------------------------------------------------------
by fabpot at 2011/08/29 03:47:50 -0700
The goal is to ensure that the load/dump calls are idempotent.
---------------------------------------------------------------------------
by michelsalib at 2011/08/29 03:56:12 -0700
Ye. Actually the load is using referenc files from the fixtures directory. My new tests will be using the dumper with the same files. Isn't that enough ? I try to stay DRY on this one.
---------------------------------------------------------------------------
by michelsalib at 2011/08/29 05:09:52 -0700
I just add unit tests and squash the commits.
I still need to be conviced about the load/dump/load unit tests. But if I did not made a point, I'll do as requested on next answer.
Commits
-------
723cb71 [Translation] Add compatibility to PCRE 6.6.0 for explicit interval pluralization
Discussion
----------
[Translation] Add compatibility to PCRE 6.6.0 for explicit interval pluralization
[Translation] Add compatibility to PCRE 6.6.0 for explicit interval pluralization
Corrected branches: From 2.0 to 2.0
more info: https://github.com/symfony/symfony/pull/2038
Commits
-------
cc098a3 [HttpKernel] Add support for xdebug.file_link_format to Debug\ExceptionHandler.php
Discussion
----------
[HttpKernel] Add support for xdebug.file_link_format to Debug\ExceptionHandler
Format file and line as url, if xdebug.file_link_format is set. Inspired by #1893
Commits
-------
020fa51 [RedirectResponse] Added missing `doctype` and `title` tag
Discussion
----------
[RedirectResponse] Added missing `doctype` and `title` tag
Commits
-------
ea0db2d [HttpFoundation] Remove useless ContentTypeMimeTypeGuesser
Discussion
----------
[2.1] [HttpFoundation] Remove useless ContentTypeMimeTypeGuesser
`mime_content_type` exists just for the compat between the old PHP 5.2
`mime_magic` extension and `file_info` extension
---------------------------------------------------------------------------
by fabpot at 2011/08/19 05:31:25 -0700
I will merge it in 2.1 as some people might rely on it.
---------------------------------------------------------------------------
by stealth35 at 2011/08/19 05:46:02 -0700
ok in the meantime, we can invert the guesser checker :
```php
/**
* Registers all natively provided mime type guessers
*/
private function __construct()
{
if (FileBinaryMimeTypeGuesser::isSupported()) {
$this->register(new FileBinaryMimeTypeGuesser());
}
if (FileinfoMimeTypeGuesser::isSupported()) {
$this->register(new FileinfoMimeTypeGuesser());
}
if (ContentTypeMimeTypeGuesser::isSupported()) {
$this->register(new ContentTypeMimeTypeGuesser());
}
}
```
---------------------------------------------------------------------------
by stloyd at 2011/08/19 05:48:38 -0700
@stealth35 You should make new PR for change you mentioned above.
---------------------------------------------------------------------------
by stealth35 at 2011/08/19 05:53:12 -0700
@stloyd done PR #1989
EDIT : forget this
Commits
-------
007e395 do not set a default CONTENT_TYPE for PATCH
fa2c027 Added support for the PATCH method
Discussion
----------
[2.1] [HttpFoundation] Added support for the PATCH method
http://tools.ietf.org/html/rfc2068#section-19.6.1.1http://tools.ietf.org/html/rfc5789
---------------------------------------------------------------------------
by Seldaek at 2011/08/07 03:23:20 -0700
According to the spec it seems that PATCH requests shouldn't be of application/x-www-form-urlencoded content-type so it shouldn't match the first if, and in the second it's probably wrong to default to application/x-www-form-urlencoded, no?
---------------------------------------------------------------------------
by lsmith77 at 2011/08/07 03:31:48 -0700
Hmm you are right. I assumed the diff would be encoded as ``application/x-www-form-urlencoded`` but there indeed is no indication of that in the spec. But given that the second case would still need some sort of handling for PATCH, just not sure what exactly ``$defaults['CONTENT_TYPE']`` should be set to.
---------------------------------------------------------------------------
by Seldaek at 2011/08/07 03:48:53 -0700
As I understand it, a PATCH request must specify a content-type or it's invalid, so we could just skip the second behavior if no content-type is present.
As your first link says:
The list of differences is in a format defined by the media type of the entity (e.g.,
"application/diff") and MUST include sufficient information to allow
the server to recreate the changes necessary to convert the original
version of the resource to the desired version.
Sounds like PATCH is highly application specific, and not so standardized, probably because it's not very useful for most purposes.
---------------------------------------------------------------------------
by lsmith77 at 2011/08/07 04:02:43 -0700
Yes, but to me this means that the patch is actually correct aside from the fact that its setting a default Content-Type, which I just corrected (not sure if this use of switch is ok with our coding style). Now if the Content-Type does end up being ``application/x-www-form-urlencoded`` then I would say its correct to decode it.
Commits
-------
24bacdc Ignore VCS files in assets:install command (closes#2025)
Discussion
----------
Ignore VCS files in assets:install command (closes#2025)
---------------------------------------------------------------------------
by stloyd at 2011/08/25 06:10:22 -0700
`ignoreVCS` is set to `true` by default, AFAIK also `getIterator()` is not needed.
---------------------------------------------------------------------------
by jalliot at 2011/08/25 06:30:32 -0700
@stloyd I knew about ``ignoreVCS`` defaulting to ``true`` but I thought it made it clearer but you're right it's not really useful.
As for ``getIterator`` I thought the conversion couldn't be made automatically on a method call like here but apparently it works so I changed it.
Thanks.
---------------------------------------------------------------------------
by tiagojsag at 2011/08/25 08:41:02 -0700
This approach creates another problem: the already existing VCS files are deleted when the command is executed, which makes at least SVN throw errors.
---------------------------------------------------------------------------
by jalliot at 2011/08/25 08:50:55 -0700
@tiagojsag If you remove the call to ``remove`` on line 83, does everything work?
Because I'm not really sure we need to remove the entire dir first since ``mirror`` should be able to adapt itself.
BTW, wouldn't it be better if you didn't commit the ``web/bundles`` dir in your SVN and instead ask to call the ``assets:install`` command each time?
---------------------------------------------------------------------------
by stof at 2011/08/25 08:58:16 -0700
Great news about SVN: the incoming 1.7 version stops adding a ``.svn`` folder in every directory but uses a single one at the root of the project (like git does for instance), solving this sort of issues about copying files :)
@tiagojsag this command has always removed the old asset folders before copying the new ones, and there is not real mean to do otherwise by keeping things simple. You could consider ignoring the ``vendor/bundles`` folder in the SVN and running the command when doing checkout (thus allowing devs to use symlinks if they want)
---------------------------------------------------------------------------
by tiagojsag at 2011/08/25 09:01:39 -0700
yes, that was the solution I was using before submitting this bug report. I also agree that it's the simplest and fastest way to address this, provided that docs get updated, so that no one spends their time trying to figure out why files are not synced with their svn repo.
---------------------------------------------------------------------------
by jalliot at 2011/08/25 09:03:11 -0700
@stof That's really great to hear!
But still this PR should be merged to avoid legacy files from current versions of SVN or other VCS.
---------------------------------------------------------------------------
by stof at 2011/08/25 09:04:31 -0700
@jalliot sure. My comment was mainly about the opposite issue raised by @tiagojsag