Commits
-------
060d75d [Console] fixed CS and simplified code
a5ff635 [Console] added support for interactive shell without readline extension
Discussion
----------
[2.1] [Console] added support for interactive shell without readline extension
---------------------------------------------------------------------------
by lmcd at 2011/06/30 15:12:56 -0700
Very useful! Pre-installed PHP on OS X ships without readline.
---------------------------------------------------------------------------
by stof at 2011/06/30 15:24:47 -0700
@lmcd You can already use the console, just not the shell.
---------------------------------------------------------------------------
by lmcd at 2011/06/30 15:27:00 -0700
@stof Yep I know, but it would be nice to use the shell :)
---------------------------------------------------------------------------
by hason at 2011/07/01 01:54:34 -0700
@stof Interactive generators will not work on Windows and Mac OS X without these simple changes. I think this is bad news.
---------------------------------------------------------------------------
by Seldaek at 2011/07/01 01:58:59 -0700
Yup, it's definitely a good addition.
---------------------------------------------------------------------------
by Seldaek at 2011/07/01 01:59:26 -0700
@fabpot: I'm not sure why this is tagged 2.1, but IMO this should go in now.
---------------------------------------------------------------------------
by stof at 2011/07/01 01:59:44 -0700
@hason This is not true. The shell is only used when running the console with the ``-s`` option. A command can be interactive without using the shell (I'm on Windows)
---------------------------------------------------------------------------
by lmcd at 2011/07/01 02:08:17 -0700
@stof Yea true, the interactive generators are fine without readline, however I don't think this dependancy should prevent people using the shell on Win/Mac.
---------------------------------------------------------------------------
by hason at 2011/07/01 02:28:10 -0700
@stof This is good news :) I was hasty.
---------------------------------------------------------------------------
by lmcd at 2011/07/04 02:03:36 -0700
I'm feeling inclined to remove readline entirely for 2.1 and roll our own solution to support things like this in the shell: http://lmcd.me/v/sf-autocomplete-2.mov
---------------------------------------------------------------------------
by stof at 2011/09/04 04:55:37 -0700
what is the status of this PR ?
Commits
-------
d675c28 [FrameworkBundle] Use Router instead of RouterInterface
ae7ae8d [FrameworkBundle] Moved router_listener from web to router.xml since it depends on the router
35a9023 [FrameworkBundle] Added isEnabled to Router commands, fixes#1467536d979 [Console] Added Command::isEnabled method that defines whether to add the command or not
Discussion
----------
[2.1] [Console] Added Command::isEnabled method
This addresses #1467.
The idea is to allow commands to evaluate whether they can run or not, since they are automatically registered.
- It's useful for the two router:* commands since they're optional (router can be disabled), but part of the FrameworkBundle that is not really optional.
- It could be useful for third party code as well.
- It's BC.
- aa95bb0d395810b29a3e654673e130736d9d1080 should address the issue in #1467, while the other commits just make sure the command is not registered at all if the router isn't standard.
One issue remains though:
- A few other services like twig helpers get the `ròuter` injected, this means that if there is really **no** router service defined, there is still an error. I'm not sure how to fix those beyond adding `on-invalid="null"` but I'm not sure if that's desirable. I guess we could argue that the router is a big candidate for replacement/suppression, and as such it should be truly optional, but if we do it I don't know where it'll lead. I don't want to end up in a situation where half the dependencies are optional to support every possible combination. @fabpot wdyt?
---------------------------------------------------------------------------
by kriswallsmith at 2011/06/28 16:19:46 -0700
I'd rather see us not register a command instead of register and then disable it. Can we do the same thing you've done here in the bundle's registerCommands() method?
---------------------------------------------------------------------------
by Seldaek at 2011/06/28 16:51:36 -0700
Note that it's never really registered. During the registration it's checked and skipped if not enabled.
However, doing it as you suggest means overriding/copy-pasting all the code from the core Bundle class, which I don't like so much. It also means adding code specific to those two commands in a somewhat unrelated place, which I also don't like.
I'm not saying the current solution is perfect, but from the alternatives I considered, it's the best I have found.
---------------------------------------------------------------------------
by stof at 2011/09/04 04:58:04 -0700
@Seldaek your branch conflicts with master. could you rebase it ?
@fabpot what do you think about this PR ?
---------------------------------------------------------------------------
by Seldaek at 2011/09/04 08:39:05 -0700
Rebased
* 2.0:
[HttpKernel] fixed typo
fixed previous merge, done the same change to other occurences
fixes usage of mb_*
Profiler session import fixed.
[Process] workaround a faulty implementation of is_executable on Windows
Swedish translation fix.
[Locale] Fix#2179 StubIntlDateFormatter support yy format
Fixed fourth argument of Filesystem->mirror()
Check that $mode in InputArgument::__construct() is not below 1 to actually look if any of the flags are set.
Check that $mode in InputOption::__construct() is not below 1 to actually look if any of the flags are set.
Check for the correct parameter type, as in InputOption (integer).
InputArgumentTest: Added test for negative integer $mode parameter input in constructor.
InputOptionTest: Added test for negative integer $mode parameter input in constructor.
Problem with fgets is that false means two things: an error or the end of the stream.
That's ok for STDIN, but it becomes a problem when using another stream (in a unit test for instance).
* The command names have now full support for nested namespaces. It means
that abbreviations work for each sub-namespace:
./app/console doctrine:mapping:info
# worked before
./app/console doctrine:map:in
# works now
./app/console doc:map:in
* Aliases are now first class citizen. They can have their own namespace,
like the main name. So, now, there is no difference between an alias and a
name.
* As names and aliases can be namespaced, the Command::getFullName() and
Command::getNamespace() method have been removed.
The current code was broken when a style was defined inline:
<bg=black>Foo</bg=black>
When creatin a new style formatter, it's better to let the formatter
apply the style to the text.
* everzet/console-privatisation-errors:
if we moved definition under private area, than we need to use only public API for that, to be able to redefine it is subclasses
start => finish, begin => end. Mixing them is a bad choice.
ability to define custom regex's for style placeholders
we need ability to get style parameters same way we set them with setStyle
sometimes, developers like me want to redefine style placeholders
in their applications. From this patch, they can acchieve this from
simple get...Regex() method redefine.
from last privatisation update $styles now private. So,
user can set style with public setter (setStyle), but can't
get this value later, which is very stupid behavior!
To keep language consistent, three methods were changed in InputOption:
* `InputOption::acceptParameter()` -> `InputOption::acceptValue()`
* `InputOption::isParameterRequired()` -> InputOption::isValueRequired()`
* `InputOption::isParameterOptional()` -> `InputOption::isValueOptional()`
The InputDefinition::asXml() method was also modified to update the `accept_value` and `is_value_required` attributes.
When an object has a "main" many relation with related "things" (objects,
parameters, ...), the method names are normalized:
* get()
* set()
* all()
* replace()
* remove()
* clear()
* isEmpty()
* add()
* register()
* count()
* keys()
The classes below follow this method naming convention:
* BrowserKit\CookieJar -> Cookie
* BrowserKit\History -> Request
* Console\Application -> Command
* Console\Application\Helper\HelperSet -> HelperInterface
* DependencyInjection\Container -> services
* DependencyInjection\ContainerBuilder -> services
* DependencyInjection\ParameterBag\ParameterBag -> parameters
* DependencyInjection\ParameterBag\FrozenParameterBag -> parameters
* DomCrawler\Form -> FormField
* EventDispatcher\Event -> parameters
* Form\FieldGroup -> Field
* HttpFoundation\HeaderBag -> headers
* HttpFoundation\ParameterBag -> parameters
* HttpFoundation\Session -> attributes
* HttpKernel\Profiler\Profiler -> DataCollectorInterface
* Routing\RouteCollection -> Route
* Security\Authentication\AuthenticationProviderManager -> AuthenticationProviderInterface
* Templating\Engine -> HelperInterface
* Translation\MessageCatalogue -> messages
The usage of these methods are only allowed when it is clear that there is a
main relation:
* a CookieJar has many Cookies;
* a Container has many services and many parameters (as services is the main
relation, we use the naming convention for this relation);
* a Console Input has many arguments and many options. There is no "main"
relation, and so the naming convention does not apply.
For many relations where the convention does not apply, the following methods
must be used instead (where XXX is the name of the related thing):
* get() -> getXXX()
* set() -> setXXX()
* all() -> getXXXs()
* replace() -> setXXXs()
* remove() -> removeXXX()
* clear() -> clearXXX()
* isEmpty() -> isEmptyXXX()
* add() -> addXXX()
* register() -> registerXXX()
* count() -> countXXX()
* keys()
