[cap-talk] "Same" key
Mark S. Miller
markm at cs.jhu.edu
Mon Feb 5 18:52:57 CST 2007
David Hopwood wrote:
> Jonathan S. Shapiro wrote:
>> But I think Bill Frantz's definition was closer to what people really
>> say and mean.
> You mean <http://www.eros-os.org/pipermail/cap-talk/2006-December/006721.html>?
> I thought that article was hopelessly confused. Yes, languages and operating
> systems tend to use different implementations of objects, but they are
> implementing the same concept. The level of abstraction at which there
> may (or may not) exist v-tables or data-bytes in capability representations,
> simply is not relevant to the definition of an object.
David, perhaps you missed the point of Bill's message? It wasn't prescriptive
-- he wasn't saying these various camps should look at things the way he
explains. Rather, it's descriptive and historical -- trying to explain why the
different camps are drawn to such different intuitions about how to carve
things up for descriptive purposes.
I think Bill calls it exactly right. Although both camps like to fancy that
they have abstracted away from implementation details, the imagery suggested
by each camp's canonical naive implementation story colors how they would give
a more abstract account.
When I first came across the fat-pointer idea (in the context of the Clarity
language, whose possible application to E I explain at
<http://erights.org/enative/fatpointers.html>), I found it surprising
precisely because the allocation unit was no longer the object. If fat
pointers were the canonical naive oo language implementation story, the PL
world may well have been drawn to describe things much like the OS folks do.
Charlie, you're the outlier here. AFAIK, you're the only one of us that has
consistently advocated the descriptive stance generally associated with the
other's camp. Perhaps further explanation of your perspective might help
bridge these perspective?
Text by me above is hereby placed in the public domain
More information about the cap-talk