This is a very important option which allows the cookie lifetime to be changed on migrate.
For example when a user converts from an anonymous session to a logged in session one might
wish to change from a persistent cookie to browser session (e.g. a banking application).
This commit allows applications to know certain meta-data about the session
Session storage is designed to only store some data against a session ID
so this method is necessary to be compatible with any session handler, including
native handlers.
Commits
-------
8dd2c27 [HttpFoundation] Further micro-optimization.
54c5d5e [HttpFoundation] Micro-optimisation.
Discussion
----------
[HttpFoundation] Micro-optimisation.
Ref #3729
---------------------------------------------------------------------------
by robocoder at 2012-03-30T11:45:02Z
If you pre-flip your $validOptions arrays, you can use isset() instead of in_array() in the loop.
This changes the performance from O(m * n) to O(m).
---------------------------------------------------------------------------
by drak at 2012-03-30T11:53:24Z
@robocoder What is the expense of the array_flip though?
---------------------------------------------------------------------------
by robocoder at 2012-03-30T11:56:21Z
Why would you use array_flip if the array doesn't change? Change $validOptions = array('x', 'y', ...) to $validOptions = array('x' => 0, 'y' => 0, ...), then change the in_array() to use isset().
---------------------------------------------------------------------------
by stof at 2012-03-30T11:57:08Z
@drak a loop. But it will be done only once before the other loop so it will be O(n + m) instead of O(m * n)
---------------------------------------------------------------------------
by drak at 2012-03-30T12:00:47Z
Ok :)
Commits
-------
5ae76f1 [HttpFoundation] Update documentation.
910b5c7 [HttpFoudation] CS, more tests and some optimization.
b0466e8 [HttpFoundation] Refactored BC Session class methods.
84c2e3c [HttpFoundation] Allow flash messages to have multiple messages per type.
Discussion
----------
[2.1][HttpFoundation] Multiple session flash messages
Bug fix: no
Feature addition: yes
Backwards compatibility break: yes, but this already happened in #2583. BC `Session` methods remain unbroken.
Symfony2 tests pass: yes
Fixes the following tickets: #1863
References the following tickets: #2714, #2753, #2510, #2543, #2853
Todo: -
This PR alters flash messages so that it is possible to store more than one message per flash type using the `add()` method or by passing an array of messages to `set()`.
__NOTES ABOUT BC__
This PR maintains BC behaviour with the `Session` class in that the old Symfony 2.0 methods will continue to work as before.
---------------------------------------------------------------------------
by drak at 2012-02-13T06:28:33Z
I think this is ready for review @fabpot @lsmith77
---------------------------------------------------------------------------
by lsmith77 at 2012-02-14T19:30:39Z
the FlashBag vs. AutoExpireFlashBag behavior and setup difference should probably also be explained in the upgrading log
---------------------------------------------------------------------------
by drak at 2012-02-15T04:43:14Z
@lsmith77 Those differences are explained already in the changelog
* Added `FlashBag`. Flashes expire when retrieved by `get()` or `all()`.
This makes the implementation ESI compatible.
* Added `AutoExpireFlashBag` (default) to replicate Symfony 2.0.x auto expire behaviour of messages auto expiring
after one page page load. Messages must be retrived by `get()` or `all()`.
---------------------------------------------------------------------------
by Crell at 2012-02-19T17:35:34Z
Drak asked me to weigh in here with use cases. Drupal currently has a similar session-stored-messaging system in place that I'd like to be able to replace with Flash messages. We frequently have multiple messages within a single request, however, so this change is critical to our being able to do so.
For instance, when saving an article in Drupal there is, by default, a "yay, you saved an article!" type message that gets displayed. If you also have the site configured to send email when a post is updated, you may see a "email notifications sent" message (depending on your access level). If you have a Solr server setup for search, and you're in debug mode, there will also be a "record ID X added to Solr, it should update in 2 minutes" message. And if there's a bug somewhere, you'll also get, as an error message rather than notice message, a "Oops, E_NOTICE on line 54" message.
Form validation is another case. If you have multiple errors in a single form, we prefer to list all of them. So if you screw up 4 times on a form, you may get 4 different error messages showing what you screwed up so you can fix it in one go instead of several.
Now sure, one could emulate that by building a multi-message layer on top of single-layer messages, but, really, why? "One is a special case of many", and there are many many cases where you'll want to post multiple messages. Like, most of Drupal. :-)
---------------------------------------------------------------------------
by lsmith77 at 2012-03-06T20:55:51Z
@fabpot is there any information you still need before merging this? do you want more discussion in which case you might want to take this to the mailing list ..
---------------------------------------------------------------------------
by drak at 2012-03-08T18:54:13Z
Another plus for this PR is that it requires no extra lines of code in templates etc to display the flashes, see https://github.com/symfony/symfony/pull/3267/files#diff-1
---------------------------------------------------------------------------
by drak at 2012-03-15T06:38:21Z
Rebased against current `master`, should be mergeable again..
---------------------------------------------------------------------------
by evillemez at 2012-03-17T03:08:41Z
+1 to this, I have an extended version of HttpFoundation just for this... would love to get rid of it.
Commits
-------
bd02554 [HttpFoundation] SPL IteratorAggregate+Countable on *Bags
665fdeb [HttpFoundation] SPL on ParameterBag
Discussion
----------
[HttpFoundation] SPL on ParameterBag
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
Added a couple SPL interfaces to ParameterBag, added shortcuts to working with the parameters. For example:
```php
<?php
$post = Request::createFromGlobal()->request;
echo "There are {count($post)} POST variables\n";
foreach ($post as $key => $val) {
echo "{$key}: {$val}\n";
}
```
Thoughts?
---------------------------------------------------------------------------
by stealth35 at 2012-03-07T13:09:11Z
You already have the `all` method
``` php
<?php
$post = Request::createFromGlobals()->request->all();
echo "There are ", count($post), " POST variables\n";
foreach ($post as $key => $val) {
echo "{$key}: {$val}\n";
}
```
---------------------------------------------------------------------------
by cboden at 2012-03-07T13:50:22Z
Yes, but when in the context of working with the Request object (or POST ParamegerBag), it's 1 more call and loose variable to set.
ParameterBag is a container, these common SPL interfaces give standard PHP container methods to it.
---------------------------------------------------------------------------
by lsmith77 at 2012-03-07T18:42:41Z
makes sense to me ..
---------------------------------------------------------------------------
by vicb at 2012-03-09T15:45:40Z
Probably makes sense. Could you check if any other `*Bag.php` needs to be updated so that it could ba an atomic merge.
---------------------------------------------------------------------------
by cboden at 2012-03-09T15:48:40Z
Whoops, good catch @vicb. I made a poor assumption all the *Bags extended ParameterBag, while only some do. I will post an update shortly.
Commits
-------
c4ee947 Native Redis Session Storage update
665f593 NativeRedisSessionStorage added
Discussion
----------
[HttpFoundation] Native Redis Session Storage
Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -
---------------------------------------------------------------------------
by lstrojny at 2012-03-04T23:15:43Z
Does Symfony (or any of its dependencies) has Redis support in any form whatsoever? If not this might be a good point to decide which clients to support
---------------------------------------------------------------------------
by lsmith77 at 2012-03-04T23:36:11Z
well ideally we just get this cache interface stuff done .. for this use case it would be perfect.
---------------------------------------------------------------------------
by pulzarraider at 2012-03-05T00:35:59Z
There is RedisProfilerStorage available (based on phpredis). I prefer and write code for [phpredis](https://github.com/nicolasff/phpredis).
It's recommended by [official Redis homepage](http://redis.io/clients#PHP). [In this benchmark](http://dev.af83.com/2011/01/01/which-php-library-to-use-with-redis-the-benchmark.html
) is fastest and less memory consumpting.
But if somebody prefer predis (with phpiredis), rediska or something other widely used, there are no limitations to add support of it to Symfony.
My opinion is, that the C extension should be supported at first, because of good performance and native session storage support. Redis is quite young and the process of creating PHP clients is comparable to Memcache.
There were created pure PHP Memcache clients in the past (Google found for example [this](http://www.phpclasses.org/browse/file/20284.html) and [this](http://code.blitzaffe.com/pages/phpclasses/files/memcached_client_52-12)), but they are not being used now. Everyone, who is seriously thinking about performance, is using only the C Redis/Memcache(d)/... extensions.
---------------------------------------------------------------------------
by drak at 2012-03-05T07:40:06Z
+1 on this PR. Needs a test written though.
I don't think there is any need to wait for #3493 imo. I'll deal with it if this is merged before #3493.
Are there any PHP ini settings for this for this driver or is everything via the `session.save_path` directive? (A quick look at the C code seems to indicate there are no explicit ini directives).
---------------------------------------------------------------------------
by lstrojny at 2012-03-05T12:14:34Z
@pulzarraider I don’t necessarily disagree with the usage of phpredis, I just wanted to bring up the issues of various clients and people having different preferences about them.
---------------------------------------------------------------------------
by fabpot at 2012-03-05T14:46:22Z
@pulzarraider Can you add some unit tests before I merge?
---------------------------------------------------------------------------
by pulzarraider at 2012-03-11T20:19:57Z
@drak No there are no php.ini settings. Only RedisArray has some, but it's another feature.
@fabpot I've added simple test based on other session storage tests.
I planned to create a RedisSessionStorage, too, but I have no time for it now. This can be added later in another PR as it's independent from NativeRedisSessionStorage.
---------------------------------------------------------------------------
by drak at 2012-03-12T02:21:25Z
The code looks OK to me.
---------------------------------------------------------------------------
by fabpot at 2012-03-15T06:05:27Z
#3493 has been merged now.
---------------------------------------------------------------------------
by pulzarraider at 2012-03-16T23:21:27Z
Code updated.
Commits
-------
5fa1c70 [json-response] Add a JsonResponse class for convenient JSON encoding
Discussion
----------
[json-response] Add a JsonResponse class for convenient JSON encoding
Usage example:
$data = array(user => $user->toArray());
return new JsonResponse($data);
---------------------------------------------------------------------------
by drak at 2012-02-16T11:51:11Z
@fabpot - maybe we could benefit with a bit more sub-namespacing in this component. One for Response for example and probably one for Request.
---------------------------------------------------------------------------
by Seldaek at 2012-02-16T15:07:31Z
@drak Please no. Moving the session was already a pain IMO since it was type-hinted in a few places (lack of interface, and interface doesn't include flash stuff still). Creating BC breaks just for fun like that is annoying for interop of bundles. It doesn't matter whether we have 10 or 15 classes in one directory.
---------------------------------------------------------------------------
by drak at 2012-02-17T08:33:46Z
@francodacosta The most optimal place is `__toString()`.
@Saldaek It just looks like the whole namespace is getting more cluttered. I suggest it because things like Request/Response objects are surely only going to grow over time. There is always the possibility to make BC for moved and renamed classes so there doesn't have to be any extra complications for making things look cleaner. Anyway, just a thought :-)
---------------------------------------------------------------------------
by stof at 2012-02-17T14:47:40Z
@drak Changing the namespace of a class is a BC break. The request and the response are used in many more places than the Session so it would be a real pain to update this. And the component is tagged with ``@api`` so BC breaks are forbidden without a good reason. The session refactoring was one as it was really an issue in the implementation, but simply renaming the class is not.
---------------------------------------------------------------------------
by fabpot at 2012-03-05T15:03:53Z
I'm -1 for adding this to the core. It does not add much value and why add a special response for JSON and not other formats?
---------------------------------------------------------------------------
by Seldaek at 2012-03-05T18:38:05Z
I think it's useful because it's a class we need in almost every project, and I don't think we're alone. It's super simple but makes me wonder every time why I have to recreate it. I don't want an additional bundle just for 3lines of code. Similarly I would say a JsonpResponse would be great, or maybe just an optional $callback arg to the json response to enable jsonp mode.
I just had someone ask me on irc how to do JSONP so while I think it's obvious and I'm sure you'd think that too, it obviously isn't to newcomers. The Response stuff is hidden behind those render methods & such and people don't realize they can simply subclass. If a few examples were in core it would be both helpful for learning and useful on a day to day basis.
As for other formats, well JSON is typically used nowadays, except when you want more fancy XML APIs, but for that the JMSSerializerBundle + FOSRestBundle are superior and we can't achieve such things in a few lines of code. I could also see a BinaryResponse or DownloadResponse or such that has proper "force-download" headers and accepts any binary stream, but that's another debate.
---------------------------------------------------------------------------
by dragoonis at 2012-03-05T19:43:05Z
I'm +1 for the concept but not commenting on how it should be implemented I'll leave that to other people.
Typically when you want to force a download you have to do ``content-disposition: attachment; filename="filehere.pdf"``
Modifying some response headers and the likes automatically for the user by returning a DownloadResponse object would be very handy..
I'm +1 for @Seldaek's point about examples of sub-classing for specific use cases. It will help with demonstrating how to do custom stuff the right way rather than people coming up with their own contraptions.
---------------------------------------------------------------------------
by stof at 2012-03-05T20:14:39Z
btw, regarding the BinaryResponse, there is a pending PR about it: #2606
---------------------------------------------------------------------------
by simensen at 2012-03-05T21:07:33Z
I'm +1 for providing reference implementations fo custom Response cases. I wanted to find best practices for handling JSONP requests/responses and couldn't find anything at all on the topic. I thought maybe extending Response might be useful but wasn't sure if that could be done safely or should be done at all.
---------------------------------------------------------------------------
by lsmith77 at 2012-03-05T22:28:01Z
@stof i think @drak was suggesting moving the class, but leaving an empty class extending from the new class in the old location to maintain BC
---------------------------------------------------------------------------
by stof at 2012-03-05T23:55:36Z
@lsmith77 This would force Symfony to use the BC class so that it does not break all typehints in existing code
---------------------------------------------------------------------------
by lsmith77 at 2012-03-06T00:22:15Z
BC hacks are never nice .. the goal would just be to eventually have all those classes and more importantly all new ones in a subnamespace. actually it might be easier to just leave all the classes in the old location and create new ones extending from the old ones. anyway .. personally i am also not such a big fan of these specialized responses .. but i guess i see FOSRestBundle as the alternative answer which makes me biased.
---------------------------------------------------------------------------
by Seldaek at 2012-03-06T07:57:36Z
I'm using FOSRestBundle when it's needed, but when you just have a small scale app that needs one or two json responses for specialized stuff it is slightly overkill. And again, newcomers probably won't know about it, and encouraging using it for simple use cases isn't exactly the best learning curve we can provide.
---------------------------------------------------------------------------
by COil at 2012-03-06T23:12:15Z
+1 for this. I have implemented such a function in all my sf1 projects, it will be the same for sf2.
---------------------------------------------------------------------------
by fabpot at 2012-03-15T13:22:27Z
Closing this PR in favor of a cookbook that explains how a developer can override the default Response class (this JSON class being a good example). see symfony/symfony-docs#1159
---------------------------------------------------------------------------
by Seldaek at 2012-03-15T13:25:08Z
Meh. Forcing people to copy paste code from the cookbook in every second project isn't exactly a step forward with regard to ease of use and user-friendliness.
---------------------------------------------------------------------------
by Seldaek at 2012-03-15T13:26:48Z
I mean following this logic, things like the X509 authentication should just be put in cookbooks too because almost nobody needs that. We have tons of code in the framework, I don't get the resistance with adding such a simple class which makes code more expressive.
---------------------------------------------------------------------------
by fabpot at 2012-03-15T13:53:07Z
because X509 authentication is not easy to get it right. Sending a JSON response is as simple as it can get:
new Response(json_encode($data), 200, array('Content-Type' => 'application/json'));
---------------------------------------------------------------------------
by marijn at 2012-03-15T13:54:25Z
Perhaps we need a `Symfony\Extensions\{Component}` namespace for things that don't necessarily belong in the core but are truly useful...
---------------------------------------------------------------------------
by Seldaek at 2012-03-15T14:03:40Z
I still fail to see why it doesn't belong in core.. There are tons of little helpers here and there, a base controller class made only of proxies, and then this gets turned down because it is simple to do it yourself? Sure it is simple, but it's repetitive and boring too. And while it's simple when you know your way around, some people aren't really sure how to do it.
The whole point of a framework is to avoid repetitive bullshit and be more productive. @fabpot do you have any real arguments against? I can see that you don't see a big use to it, fair enough, but do you see any downside at all?
It does not make sense to try and store session ini directives since they can be changes outside
of the class as they are part of the global state.
Coding stan
The HTTP status code translation table was updated to include all HTTP status codes as defined by the IANA Hypertext Transfer Protocol (HTTP) Status Code Registry (http://www.iana.org/assignments/http-status-codes/).
Commits
-------
4f8e8ef Improving performance on digit filtering
Discussion
----------
Improving performance on digit filtering
I haven't tested it on a productive system but I think it should be way faster to use filter_var() instead of preg_replace() for several reasons.
This is my first pull request for symfony and I don't know how you do those kind of performance tests but please verify my assumption if you can :-)
Maybe we can also use filter_var() to replace other regular expressions :-)
HTH =)
---------------------------------------------------------------------------
by drak at 2012-02-22T00:35:44Z
@Toflar - nice move +1
---------------------------------------------------------------------------
by drak at 2012-02-22T18:53:40Z
@Toflar - Maybe you can bench the changes using this as a template: https://gist.github.com/1356129
---------------------------------------------------------------------------
by Toflar at 2012-02-23T13:18:18Z
I have already. And it's way faster, otherwise I wouldn't have opened a pull request ;) But obviously it strongly depends on the length of the string and the environment. That's why I was wondering whether you have a general performance tests environment ;) Because the results strongly depend on other factors, there's - in my opinion - no point in exact results. If a general info is sufficient: my tests for the regex resulted in about 7 - 8 microseconds whereas the filter version only took 1.5 - 2 microseconds for the same string.
Commits
-------
471b564 auto_start should be false
6e2a7da Support session cookie options with cookie_ prefix
e0fba80 Properly merge session cookie_* parameters
Discussion
----------
Set session.cookie_* parameters properly
Bug fix: yes
Feature addition: no
Backwards compatibility break: yes
Symfony2 tests pass: yes
Fixes the following tickets: /
Cookie parameters in $options are not prefixed with cookie_ the same is true for data returned from session_get_cookie_params.
I've marked this as BC because the options that get dumped into the container have different name. But I don't think anybody was actually changing them or accessing them in their bundles.
P.S. @drak also desires some credits for this PR as I incorporated some lines written by him in one of the iterations.
---------------------------------------------------------------------------
by drak at 2012-02-23T14:24:42Z
@mvrhov - what does this fix exactly? It looks like a different way of doing the same thing but now there is no default value on `cookie_httponly`.
---------------------------------------------------------------------------
by mvrhov at 2012-02-23T15:09:17Z
Like I said in description. $option contains some cookie options and none of them has cookie_ prefix.
And this prefix is needed in two cases:
- to properly merge defaults and override them with what user set
- in a foreach for for proper ini_set
Sorry non native speaker an a bit hard to explain, could you ping me in a couple of hours on IRC if this still doesn't make any sense.
---------------------------------------------------------------------------
by drak at 2012-02-23T15:29:41Z
@mvrhov - I wrote some tests for this particular code and I still don't see what this PR fixes. I'll try to catch you on IRC later on but can't guarantee it.
---------------------------------------------------------------------------
by mvrhov at 2012-02-23T16:02:41Z
added test
---------------------------------------------------------------------------
by drak at 2012-02-24T08:30:51Z
Just for reference for those reading this ticket, `session_set_cookie_params()` alters the runtime ini settings it corresponds to see http://docs.php.net/manual/en/function.session-set-cookie-params.php so we agreed to remove the special handling that was present since it is redundant.
---------------------------------------------------------------------------
by dlsniper at 2012-02-28T22:19:32Z
Hi, Is this patch relevant or not after all?
ping @drak @mvrhov
Thanks :)
---------------------------------------------------------------------------
by drak at 2012-02-29T03:34:22Z
It is relevant. Maybe I'll do the cleanup this PR by forking it if @mvrhov doesn't have time.
---------------------------------------------------------------------------
by mvrhov at 2012-02-29T05:40:47Z
Fixed the typo and changed the false to ture as reported in comments. I've also rebased. I'll see what I can do about config file change later today. Sorry for the delay, been too busy for the past week.
---------------------------------------------------------------------------
by mvrhov at 2012-02-29T08:49:23Z
I've also done the config part.
---------------------------------------------------------------------------
by mvrhov at 2012-02-29T11:01:14Z
Ok, this should be it.
---------------------------------------------------------------------------
by drak at 2012-03-01T00:59:16Z
@fabpot - looks good from my side.
Added blocks, updated links and references and fixed typos.
Note it is not possible to throw exceptions in the write or close methods of a session save handler.
Commits
-------
bafcaaf Removed version field
f9d9dc7 Add branch-alias for composer
Discussion
----------
Add branch-alias for composer
This should restore the 2.1-dev version (as an alias of dev-master) so that `2.*` or `2.1.*` constraints work again. I'll adjust packagist soon to also display those aliases.
Commits
-------
fb2bb65 [HttpFoundation] Fix session.cache_limiter is not set correctly
Discussion
----------
[HttpFoundation] Fix session.cache_limiter is not set correctly
Bug fix: yes
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -
Fixes a regression after the session refactoring where extra cache control http headers are sent.
This was previously handled by [calling session_cache_limiter(false) in NativeSessionStorage](https://github.com/symfony/symfony/blob/2.0/src/Symfony/Component/HttpFoundation/SessionStorage/NativeSessionStorage.php#L81)
---------------------------------------------------------------------------
by drak at 2012-02-21T12:23:48Z
@fabpot - this code can be merged imo.
Commits
-------
d077ede [HttpFoundation] Increase test coverage.
cbb3e69 [HttpFoundation] Increase test coverage.
Discussion
----------
[HttpFoundation] Increase session test coverage.
Bug fix: no
Feature addition: no
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -
Rename ArraySessionStorage to make it clear the session is a mock for testing purposes only.
Has BC class for ArraySessionStorage
Added sanity check when starting the session.
Fixed typos and incorrect php extension test method
session_module_name() also sets session.save_handler, so must use extension_loaded() to check if module exist
or not.
Respect autostart settings.
Session object now implements SessionInterface to make it more portable.
AbstractSessionStorage and SessionSaveHandlerInterface now makes implementation
of session storage drivers simple and easy to write for both custom save handlers
and native php save handlers and respect the PHP session workflow.
This commit outsources the flash message processing to it's own interface.
Overall flash messages now can have multiple flash types and each type can
store multiple messages. For convenience there are now four flash types
by default, INFO, NOTICE, WARNING and ERROR.
There are two concrete implementations: one preserving the old behaviour of
flash messages expiring exactly after one page load, regardless of being
displayed or not; and the other where flash messages persist until explicitly
popped.
This commit outsources session attribute storage to it's own class.
There are two concrete implementations, one with structured namespace storage and the other
without.
Apache expects the response to already be in chunked format in that case,
which causes it to not deliver the streamed body.
If no Content-Length is set on the response, web servers will automatically
switch to chunked Transfer-Encoding, and handle the chunking for you.
Nginx does not share the issue that apache has, but will add the Content-
Length header too.
Commits
-------
887c0e9 moved EngineInterface::stream() to a new StreamingEngineInterface to keep BC with 2.0
473741b added the possibility to change a StreamedResponse callback after its creation
8717d44 moved a test in the constructor
e44b8ba made some cosmetic changes
0038d1b [HttpFoundation] added support for streamed responses
Discussion
----------
[HttpFoundation] added support for streamed responses
To stream a Response, use the StreamedResponse class instead of the
standard Response class:
$response = new StreamedResponse(function () {
echo 'FOO';
});
$response = new StreamedResponse(function () {
echo 'FOO';
}, 200, array('Content-Type' => 'text/plain'));
As you can see, a StreamedResponse instance takes a PHP callback instead of
a string for the Response content. It's up to the developer to stream the
response content from the callback with standard PHP functions like echo.
You can also use flush() if needed.
From a controller, do something like this:
$twig = $this->get('templating');
return new StreamedResponse(function () use ($templating) {
$templating->stream('BlogBundle:Annot:streamed.html.twig');
}, 200, array('Content-Type' => 'text/html'));
If you are using the base controller, you can use the stream() method instead:
return $this->stream('BlogBundle:Annot:streamed.html.twig');
You can stream an existing file by using the PHP built-in readfile() function:
new StreamedResponse(function () use ($file) {
readfile($file);
}, 200, array('Content-Type' => 'image/png');
Read http://php.net/flush for more information about output buffering in PHP.
Note that you should do your best to move all expensive operations to
be "activated/evaluated/called" during template evaluation.
Templates
---------
If you are using Twig as a template engine, everything should work as
usual, even if are using template inheritance!
However, note that streaming is not supported for PHP templates. Support
is impossible by design (as the layout is rendered after the main content).
Exceptions
----------
Exceptions thrown during rendering will be rendered as usual except that
some content might have been rendered already.
Limitations
-----------
As the getContent() method always returns false for streamed Responses, some
event listeners won't work at all:
* Web debug toolbar is not available for such Responses (but the profiler works fine);
* ESI is not supported.
Also note that streamed responses cannot benefit from HTTP caching for obvious
reasons.
---------------------------------------------------------------------------
by Seldaek at 2011/12/21 06:34:13 -0800
Just an idea: what about exposing flush() to twig? Possibly in a way that it will not call it if the template is not streaming. That way you could always add a flush() after your </head> tag to make sure that goes out as fast as possible, but it wouldn't mess with non-streamed responses. Although it appears flush() doesn't affect output buffers, so I guess it doesn't need anything special.
When you say "ESI is not supported.", that means only the AppCache right? I don't see why this would affect Varnish, but then again as far as I know Varnish will buffer if ESI is used so the benefit of streaming there is non-existent.
---------------------------------------------------------------------------
by cordoval at 2011/12/21 08:04:21 -0800
wonder what the use case is for streaming a response, very interesting.
---------------------------------------------------------------------------
by johnkary at 2011/12/21 08:19:48 -0800
@cordoval Common use cases are present fairly well by this RailsCast video: http://railscasts.com/episodes/266-http-streaming
Essentially it allows faster fetching of web assets (JS, CSS, etc) located in the <head></head>, allowing those assets to be fetched as soon as possible before the remainder of the content body is computed and sent to the browser. The end goal is to improve page load speed.
There are other uses cases too like making large body content available quickly to the service consuming it. Think if you were monitoring a live feed of JSON data of newest Twitter comments.
---------------------------------------------------------------------------
by lsmith77 at 2011/12/21 08:54:35 -0800
How does this relate the limitations mentioned in:
http://yehudakatz.com/2010/09/07/automatic-flushing-the-rails-3-1-plan/
Am I right to understand that due to how twig works we are not really streaming the content pieces when we call render(), but instead the entire template with its layout is rendered and only then will we flush? or does it mean that the render call will work its way to the top level layout template and form then on it can send the content until it hits another block, which it then first renders before it continues to send the data?
---------------------------------------------------------------------------
by stof at 2011/12/21 09:02:53 -0800
@lsmith77 this is why the ``stream`` method calls ``display`` in Twig instead of ``render``. ``display`` uses echo to print the output of the template line by line (and blocks are simply method calls in the middle). Look at your compiled templates to see it (the ``doDisplay`` method)
Rendering a template with Twig simply use an output buffer around the rendering.
---------------------------------------------------------------------------
by fabpot at 2011/12/21 09:24:33 -0800
@lsmith77: We don't have the Rails problem thanks to Twig as the order of execution is the right one by default (the layout is executed first); it means that we can have the flush feature without any change to how the core works. As @stof mentioned, we are using `display`, not `render`, so we are streaming your templates for byte one.
---------------------------------------------------------------------------
by fabpot at 2011/12/21 09:36:41 -0800
@Seldaek: yes, I meant ESI with the PHP reverse proxy.
---------------------------------------------------------------------------
by fabpot at 2011/12/21 09:37:34 -0800
@Seldaek: I have `flush()` support for Twig on my todo-list. As you mentioned, It should be trivial to implement.
---------------------------------------------------------------------------
by fzaninotto at 2011/12/21 09:48:18 -0800
How do streaming responses deal with assets that must be called in the head, but are declared in the body?
---------------------------------------------------------------------------
by fabpot at 2011/12/21 09:52:12 -0800
@fzaninotto: What do you mean?
With Twig, your layout is defined with blocks ("holes"). These blocks are overridden by child templates, but evaluated as they are encountered in the layout. So, everything works as expected.
As noted in the commit message, this does not work with PHP templates for the problems mentioned in the Rails post (as the order of execution is not the right one -- the child template is first evaluated and then the layout).
---------------------------------------------------------------------------
by fzaninotto at 2011/12/21 10:07:35 -0800
I was referring to using Assetic. Not sure if this compiles to Twig the same way as javascript and stylesheet blocks placed in the head - and therefore executed in the right way.
---------------------------------------------------------------------------
by fabpot at 2011/12/21 10:34:59 -0800
@Seldaek: I've just added a `flush` tag in Twig 1.5: 1d6dfad4f5
---------------------------------------------------------------------------
by catchamonkey at 2011/12/21 13:29:22 -0800
I'm really happy you've got this into the core, it's a great feature to have! Good work.