Re: Taming Java Libraries in E vs. Joe-e

On Nov 14, 2008, at 15:18, Jimmy Wylie wrote:

> It seems E may use it in its implementation of its own boolean type,  
> but it may also add or remove things, making a simple comparison  
> between safej files incomplete.

Look at Sugarings in jsrc/org/erights/e/elib/prim/ScriptMaker.java to  
find the classes which add things. This information *should* be in the  
safej table but it isn't yet.

> I did actually look at Boolean.safej.  In both, the method  
> getBoolean()
> was suppressed because it accesses a system property.  Joe-E allows  
> more
> methods than E it seems, like toString() and parseBoolean(), but those
> seem like methods E wouldn't have to tame because they would be
> implemented elsewhere in the E language.

getBoolean() is probably suppressed because it would be silly in E - E  
does not distinguish boxed and unboxed booleans.

toString is universally suppressed for several reasons:
   - the default implementation of it exposes pointer bits  
(nondeterminism)
   - all the rest of java.lang.Object's protocol, including toString,  
is suppressed
   - E has its own to-string mechanism (E.toString) which has cycle  
detection and also uses Java classes' toString when appropriate

(It's been a while since I looked at this stuff; I may be a little  
off. MarkM?)

> Also, I do have one other question about the safej files.  In Joe-E,  
> they use the static tag and instance tag.  In E's safej files, they  
> use a statics tag and methods tag.  Are the methods tag and instance  
> tag denoting the same methods?

I don't know Joe-E, but probably yes.

>   Or is E using that tag to add its own methods?  I noticed in  
> Boolean.safej that under "methods", E has listed some methods I  
> couldn't find in the Java API like and(), not(), or(), xor(), and  
> pick(obj, obj).

These are methods added by E (sugar classes, which are defined in a  
table in SafeJ.java). safej files control methods *after* the E  
property-mangling, name-mangling and sugaring stages have been  
applied. (IMO this should not be the case, but I'm not an expert on  
that aspect of the system.)

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>