Re: Embed E in Java

On Tue, Jan 12, 2010 at 2:13 PM, Kevin Reid <[email protected]> wrote:

> On Jan 12, 2010, at 17:08, Mark Miller wrote:
> > On Tue, Jan 12, 2010 at 1:42 PM, Kevin Reid <[email protected]> wrote:
> >> On Oct 21, 2009, at 13:25, Mark Miller wrote:
> >>> On Tue, Oct 20, 2009 at 9:17 AM, Kevin Reid <[email protected]> wrote:
> >>>> MarkM: Is there a non-historical reason NestedException/
> >>>> NestedThrowable/etc exist instead of using the Java 1.4
> >>>> Throwable.getCause() mechanism?
> >>>
> >>> Mostly history. The NestedException,.. mechanism predates Java 1.4.
> >>> Since it worked, I have never examined the new getCause() API. I do
> >>> not know whether or not it would be safe to use instead.
> >>
> >> I believe I understand it sufficiently well to say that it is safe
> >> and
> >> appropriate to use instead. I propose to change E-on-Java to replace
> >> the Nested* mechanisms (backwards compatibly, i.e. the .leaf() sugar
> >> and so on will still exist) with causes.
> >>
> >> * What is the minimum Java version E-on-Java is intended to require?
> >>
> >>
> > 1.4. Yes, I realize that contradicts my paragraph quoted above. I
> > have no
> > objections to moving E-on-Java's dependencies forward to 1.5. Does
> > anyone
> > else see any reasons to stay with 1.4 for now?
>
> Hold on. Where did 1.5 come in, and what do you mean by "moving E-on-
> Java's dependencies"? I'm only proposing to use 1.4 features, not 1.5.
>
> I don't see any contradiction in what you have said so far.
>
> My mistake. I assumed (incorrectly) that getCause() came in with 1.5.



> > Btw, one big potential benefit of moving towards 1.5 is that we
> > could start
> > making parts of E-on-Java be Joe-E safe. Ultimately, except for a
> > small
> > powerbox, it might be good for E-on-Java to become E-on-Joe-E.
> > Unfortunately, for this to happen, we'd need to start to use Java's
> > generics
> > (parameterized typed) within the E implementation, at an enormous
> > cost in
> > readability and reviewability. I do not yet understand how practical
> > it is
> > to use Joe-E while avoiding generics.
>
> Why do you consider the generics an enormous cost? Additional
> verbosity, or erasure-related issues, or what?
>
> Both. When participating in the Waterken/Joe-E security review, the effort
to ignore all the information (or worse, *almost* all the information)
between angle brackets, because it could be lying without warning, made
everything much harder. For this noise to become a large fraction of the
total source text is more than annoying. All this is Java 1.5's fault, not
Joe-E's. Joe-E was not in a practical position to repair this mistake.



> --
> Kevin Reid                                  <http://switchb.org/kpreid/>
>
>
>
>
> _______________________________________________
> e-lang mailing list
> [email protected]
> http://www.eros-os.org/mailman/listinfo/e-lang
>



-- 
Text by me above is hereby placed in the public domain

   Cheers,
   --MarkM

_______________________________________________
e-lang mailing list
[email protected]
http://www.eros-os.org/mailman/listinfo/e-lang