Re: Progress with CGI::Session::Serialize::yaml

On Thu, Feb 23, 2006 at 03:00:25PM -0800, Tyler MacDonald wrote:
>       I've played around with the four YAML modules people seem most
> interested in (YAML, YAML::Syck, JSON, JSON::Syck). 
> 
>       - YAML passes all the existing serialization unit tests
> 
>       - YAML::Syck passes all except the last one (101 - Overloaded
> objects have matching addresses)
> 
>       - JSON and JSON::Syck fail to rebless references back into their
> original package, but this is pretty much expected behaviour.
> 
>       What do you think is the best way to move forward? Cosidering all
> those serializers have the same interface, I was thinking of having ::yaml
> default to the regular YAML package, and then if people want they can
> override that with something like
> 
>       local $CGI::Session::Serializa::yaml::FLAVOUR = "JSON::Syck";
> 
>       ... I would really like to default to YAML::Syck instead since it's
> much much faster, but it doesnt appear to deal with duplicate references as
> well as YAML.

I think there are reasons for people to want to use all the different
drivers. We discussed JSON vs. YAML previously, and ::Syck will be
preferred for performance when a C compiler is available, otherwise the
pure Perl versions will be desired. 

I'll throw out this:

 - document the limitations of each clearly
 - adjust the tests to match the limitations of each.

For example, I know I want to keep using YAML::Syck, and I don't care
about "overloaded objects havin matching addresses".  

Other perspectives?

        Mark


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642