From: Chris Morris <chrismo@...>
Date: 2002-03-19T09:51:43+09:00
Subject: RE: [ANN] Xml Serialization for Ruby

> I agree that that smells ... the inconsistency of "root element is class
> name, all other nodes are instance var names" ... but I'm almost positive
> this is how C# does it, and when I was working with it felt quite natural
> from a 'user' standpoint. The inconsistency never occurred to me
> until now.
> Of course, I need to double check -- work has detoured me from my
> C# lately.

More on this. If I have this class:

  class Person
    attr_accessor :name, :age, :height

    ...
  end

I'd rather see this xml:

  <Person>
    <name>Chris Morris</name>
    <age>30</age>
    <height>72</height>
  </Person>

... than this:

  <Person>
    <String name="name">Chris Morris</String>
    <Integer name="age">30</Integer>
    <Integer name="height">72</Integer>
  </Person>

... because my first thought is -- "Why doesn't the xml look like the class
declaration? Where'd all these types come from? -- I don't care about that
so much as putting in the data. I thought this was Ruby." :)

I don't think -- "hey, that first xml is inconsistently structured" (even
though it is) ... To me, the first structure comes off very much like I
typed up the Ruby class. And the first structure is faster to grok -- my eye
tends to brush over attributes at first shot. I think it has better PoLS.

But again, I guess I shouldn't add any fuel to this fire, because I'm
willing to support multiple formats; it's simple enough to do and I'd hate
to have a tool like this work great for small stuff, than fail just because
I need to ramp it up a bit in the typing department -- plus for me to go
with a strictly typeless xml option, I have to have a fully initialized
object, and I don't want to force that on anyone either -- give them the
option of type info in the xml or a fully initialized object.

Chris