Eventmachine talk - Nov 2007 - For Tony Arcieri, on EM::Connection#post_init

Tony, the mods you checked in pass all the standard tests, but I hit a
regression with my REST framework. It assumes that #post_init will be called
inside of the EM::Connection base class initializer. I had something like
this:

class Something < EM::Connection
  def initialize *args
    super
    p @my_ivar
  end

  def post_init
    @my_ivar = :something
  end

end


It was actually a lot more complex because the #post_init was written in C
(it was in an extension) but the point is the same. With your new code, this
example will give an error.

I''m not terribly concerned about this regression because I suspect it
won''t
affect too much code, but we should be clear on whether the documented
behavior has changed.

Can you give a concise description of how #initialize and #post_init
interact, including the role of the newly-supported parameters, so we can
publish it and know what to expect? Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/eventmachine-talk/attachments/20071102/e8698211/attachment.html

#post_init is now being called within EM::Connection.new

It''s occurring after both allocate and initialize, but before EM::
Connection.new returns a new instance of an EM::Connection object.

#initialize is effectively a stub in the base class now.  That way calling
super is optional (and a noop)

Because of that, #post_init will always be called only after #initialize
returns

On 11/2/07, Francis Cianfrocca <garbagecat10 at gmail.com>
wrote:
>
> Tony, the mods you checked in pass all the standard tests, but I hit a
> regression with my REST framework. It assumes that #post_init will be
called
> inside of the EM::Connection base class initializer. I had something like
> this:
>
> class Something < EM::Connection
>   def initialize *args
>     super
>     p @my_ivar
>   end
>
>   def post_init
>     @my_ivar = :something
>   end
>
> end
>
>
> It was actually a lot more complex because the #post_init was written in C
> (it was in an extension) but the point is the same. With your new code,
this
> example will give an error.
>
> I''m not terribly concerned about this regression because I suspect
it
> won''t affect too much code, but we should be clear on whether the
documented
> behavior has changed.
>
> Can you give a concise description of how #initialize and #post_init
> interact, including the role of the newly-supported parameters, so we can
> publish it and know what to expect? Thanks.
>
> _______________________________________________
> Eventmachine-talk mailing list
> Eventmachine-talk at rubyforge.org
> http://rubyforge.org/mailman/listinfo/eventmachine-talk
>


--
Tony Arcieri
ClickCaster, Inc.
tony at clickcaster.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/eventmachine-talk/attachments/20071102/0416b6c7/attachment-0001.html

On Fri, 2007-11-02 at 03:10 -0400, Francis Cianfrocca
wrote:
> Tony, the mods you checked in pass all the standard tests,
> but I hit a regression with my REST framework.
Hi Francis

Sorry to hijack the thread... Is this REST framework something you have
released or intend to release? Building a REST API server was the reason
I started investigating eventmachine, so if this code is publicly
available, I''d like to have a look at it :)

Best,
Andre

On 11/2/07, Andre Nathan <andre at digirati.com.br>
wrote:
>
> On Fri, 2007-11-02 at 03:10 -0400, Francis Cianfrocca wrote:
> > Tony, the mods you checked in pass all the standard tests,
> > but I hit a regression with my REST framework.
>
> Hi Francis
>
> Sorry to hijack the thread... Is this REST framework something you have
> released or intend to release? Building a REST API server was the reason
> I started investigating eventmachine, so if this code is publicly
> available, I''d like to have a look at it :)


I''ll try to get it released over the weekend.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/eventmachine-talk/attachments/20071102/49d6e7ff/attachment.html