Re: [Cgi-session-user] Re: Serialize::yaml,json

[Tyler M. <ty...@yi...>]

Mark Stosberg <ma...@su...> wrote:
> > 	JSON and YAML serializers are in trunk as of change 249. I've logged
> > into my PAUSE account and deleted the CGI::Session::Serialize::yaml
> > distribution; it is scheduled to disappear on March 10th.
> > 
> > 	What else needs to be done for the next release of CGI::Session to
> > happen?
> Thanks for this work. I'm happy to make another release with just this
> improvement, if that helps some other project progress. 
> 
> However, I also want to continue the discussions about what to do about
> "flush()" to see if we can agree on a short-term solution for that. 
> 
> Sherzord has weighed in, and I'd like to respond further on the issue.

	Sounds good. Let me know if there's anything I can do. Apache2::AUS
still needs some more test/doc/examples before I can release it, but I don't
want to release until I can use JSON serialization by default.

	I think I missed the conversation on flush(), my mailbox has been
bombarded recently and I havent looked at everything. :) One comment about
flush() tho:

	AUS's session driver uses a table format like this with additional
metadata:

CREATE TABLE aus_session (
    id          varchar(128) NOT NULL PRIMARY KEY,
    created     TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now(),
    time_last   TIMESTAMP WITH TIME ZONE NULL,
    user_id     INT NULL,
    a_session   BYTEA NULL,
    
    FOREIGN KEY (user_id)
        REFERENCES  aus_user (id)
        ON UPDATE CASCADE
        ON DELETE SET NULL
);

	In order to have the driver update user_id and time_last, I had to
modify flush() to pass $session to the driver's store() method. In order to
change that one line of functionality I had to copy-and-paste the entire
50-line flush method. I'd like to see flush() broken down into smaller bits;
maybe instead of having it call the driver and serializer itself,
CGI::Session could have it's own freeze/store/delete methods? By default
they'd just call the driver/serializer methods, but then subclassing becomes
easier.

	Just a thought...

	Cheers,
		Tyler