[ANN] Ron 0.1.0 Released

C

Caleb Clausen

Ron version 0.1.0 has been released!

http://rubyforge.org/projects/ron

Well, who needed another serialization format? Not me, but that's what
I've made. Ruby Object Notation (Ron) is a textual format for the
representation of Ruby data structures. It's somewhat like YAML, XML,
or (most of all) JSON. However, since it is defined entirely within and
as a subset of Ruby, it has the slick property that Ron expressions are
legal Ruby. Thus it is very like JSON, except that it's Ruby-centered
instead of being JavaScript-centered.

Another way to look at Ron is as a purely declarative language for
creating (almost) any type of Ruby data structure.

Ruby already has formats for representing a literal Array, Hash, Set,
String, Symbol, Integer, etc. Ron simply reuses the same notations for
those data types. The major new feature of Ron is that it provides a
literal syntax for representing arbitrary Ruby Objects.

Say you have a class that looks like this:

class K
attr :a
attr :b
end

A Ron literal that represents a K instance might look like this:

K-{:mad:a=>1, :mad:b=>2}

Better yet, if you provide setters as well as getters, you can leave
off the @ signs:

class K
attr_accessor :a,:b
end

K-{:a=>1, :b=>2}


This construct breaks encapsulation, so use the capability with
moderation and prudence, please.

An existing object can be rendered into Ron form with a call to
Ron.dump:

Ron.dump([1,2,3]) #=>"[1, 2, 3]"

Ron.load is the inverse operation; it takes a Ron-formatted string and
converts it back to a live Ruby object. (This is just an alias of
eval.):

Ron.load("K-{:mad:a=>1, :mad:b=>2}") #=> #<K:mad:a=2, @b=2>


bugs and limitations
Currently, some types of ruby object cannot be dumped. In some
cases, this is because it is impossible to (correctly) dump those
objects. In others, it's merely difficult. Here are the ones I
think I should eventually be doing, mostly with ParseTree help:

(anonymous) Class and Module
Method
UnboundMethod
Proc

These are very difficult, and I don't see how to approach them
correctly (but it would be very nice...):

Thread
Continuation

These are more or less impossible:

File
Dir
Process
IO
 
J

Jonas Pfenniger

This looks really interesting. So you overloaded the - operator on
classes to achieve the K - {} ?
 
C

Caleb Clausen


Both are ruby-based serialization packages/languages. Amarshal
generates imperative code, tho, whereas in Ron I've tried to take a
purely declarative approach.
For instance the amarshal representation of [1,2,3] (taken from the website) is:

v = []
v[0] = Array.allocate()
v[0] << 1
v[0] << 2
v[0] << 3
v[0]

In Ron, it's merely: "[1,2,3]". A more interesting example would be a
recursive data structure, say an array containing itself. Here's
amarshal (extrapolating , haven't tried this):

v = []
v[0] = Array.allocate()
v[0] << v[0]
v[0]

Versus Ron:

Recursive(v1_={}, [v1_])


Since this is essentially a declarative problem, my declarative syntax
should be shorter and (hopefully) more readable in most cases.

At first glance, the notation

K-{:a=>1, :b=>2}

for a literal instance of K seems new and interesting.

Well, thanks. As to Jonas's question about this syntax, yes, it's just
a matter of defining Class#-.
 
T

Trans

Jonas said:
This looks really interesting. So you overloaded the - operator on
classes to achieve the K - {} ?

I agree, interesting. I would recommend howerver using something other
then '-', b/c there is talk of using that for trait-like extensions to
class/module in future versions of Ruby. '<<' comes to mind a possilble
alternative.

K << { a:1, b:2 }

T.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,767
Messages
2,569,572
Members
45,045
Latest member
DRCM

Latest Threads

Top