Lukas Kahwe Smith wrote:
> Andreas Stephan wrote:
> 
>> Although it seems they did not turn the helpers off as they did with
>> cake, the results are pretty compareable to the above.
>> So I guess symfony could use some performance optimization in the next
>> major release.
> 
> I do not know these other frameworks internally at all, so I do not know 
> if they are using similar approaches. I see a few areas that I would 
> like to see investigated:
> 
> 1) filter chain

As I mentioned in a posting before: Saying the filter chain is evil is a
bit too simple..

We have to separate the problems: The filter chain is a concept that is
pretty smart IMHO. But the implementation (namely the filters) could get
some optimizations like lazy loading.

Also the sfController::forward() has much overhead that is not needed
with each request / forward.

What I see is that if we change something here we should separate
between a) request and b) module/action call

I guess separating these parts could help optimize the whole framework
and last but not least could enable us to provide a nice page controller.

My opinion is that we need the controller to work in a modular way so we
can easily turn on/off things for better performance. For example: Check
for a generator.yml, check if application is on/off. I think there is a
file system call overhead with configuration files at all, isn't it?


> 2) how does overloading affect performance
> 3) routing performance (we have a lot of flexibility, but maybe we need 
> to make it more clear what costs what)
> 4) configuration loading (sfOptimizer is a scary approach, see my recent 
>   post on symfony-dev@ on my micro benchmarks)

I wonder how we can check these things reliable.

We have the sfTimer and sfTimerManager.

So we could add some more timers to get more information.

Maybe also some sort of analyzer module that takes a deeper look into
log files to display average results over several requests and so on.

The other way is to use external tools like ab or siege which is
probably a more reliable approach but with the drawback that it is hard
to examn isolate single performance problems.

Most important is to find one or some reference configuration setup(s)
and reference module(s) to make speed tests really comparable.

Regards,
Matthias


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"symfony users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/symfony-users?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to