On 18/04/2007, at 8:25 AM, Charles Oliver Nutter wrote:

> Jon Tirsen wrote:
>> Awesome! This seems to solve the problems I was having! Our app is
>> actually now relatively stable in JRuby.
>>
>> It is certainly the correct semantics for the session wrt Rails
>> compatability. In Rails your session will always be serialized after
>> each request.
>
> I'd love to hear some perf numbers for marshalling. My last rewrite
> simplified it quite a bit, and it's all Java code, so I wouldn't  
> expect
> it to be slow, but I'd like to know if it's going to be slower or  
> faster
> than MRI. Gotta figure out how many hurdles there are left to leap  
> over.

In general Rails apps don't save that much in their sessions. I think  
we have a handful of strings and integers and the occassional array.  
Rails encourages stateless REST style server sides. IIRC I even think  
the 2.0 default session storage is client side cookies (which  
obviously has security issues, which I'm sure they've worked through).

In short, I don't think marshaling speed is going to have a big  
impact on Rails performance. (Unless it's embarrassingly slow though.)

This will obviously change if someone writes a stateful component- 
based web framework for Rails (a la Seaside/Wicket).

Cheers,
Jon




----
New from ThoughtWorks: Mingle, an Agile project management application.
Mingle. Project Intelligence. Powerfully Simple.
More at http://studios.thoughtworks.com

_______________________________________________
Jruby-extras-devel mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/jruby-extras-devel

Reply via email to