[e-lang] Freezing Kernel-E eval (was: Remaining semantic issues)

Mark Miller markm at cs.jhu.edu
Mon Aug 15 22:15:27 EDT 2005


At <http://www.eros-os.org/pipermail/e-lang/2005-January/010416.html> Kevin 
enumerated some remaining issues in settling the evaluation semantics of 
Kernel-E. Here's our status:


Kevin Reid wrote:
> This is a list of bugs which I believe will require changes to Kernel-E  
> semantics or structure.
> 
> *** SuchThatPattern discards information [...]

Closed
https://sourceforge.net/tracker/index.php?func=detail&aid=1212442&group_id=75274&atid=551529
https://sourceforge.net/tracker/index.php?func=detail&aid=1211112&group_id=75274&atid=551529


> *** Failing patterns in kernel matchers / expansion of multiple matchers [...]

Closed
https://sourceforge.net/tracker/index.php?func=detail&aid=1211088&group_id=75274&atid=551529


> *** Exception information leakage [...]

Still open
https://sourceforge.net/tracker/index.php?func=detail&aid=1211106&group_id=75274&atid=551529
https://sourceforge.net/tracker/index.php?func=detail&aid=1211104&group_id=75274&atid=551529


In addition, we have yet to settle on a practical & POLA auditing protocol.

Still open
https://sourceforge.net/tracker/index.php?func=detail&aid=1260156&group_id=75274&atid=551529
https://sourceforge.net/tracker/index.php?func=detail&aid=1260492&group_id=75274&atid=551529
https://sourceforge.net/tracker/index.php?func=detail&aid=1260491&group_id=75274&atid=551529

The text at the last of these links says:
> The current experimental E-on-Java auditing protocol relied on unshadowable
> names in a way that made it impractical. (Thanks to Kevin Reid for pointing
> this out.) Also, the 0.8.36 guard revolution removes guards from
> FinalPattern & SlotPattern, thereby making getSynEnv() useless for
> auditing.
> 
> OTOH, the current experimental E-on-CL auditing protocol violates POLA.
> It's also hard to imagine how adequately memoize it, in order to avoid
> needless per-instance overheads.
> 
> Alan Karp has suggested a way out which I will try to explain.


Since exceptions may depends on how we do auditing, but the reverse is less 
likely, I'm inclined to do auditing next and then exceptions.


What other issues remain open in the evaluation semantics of Kernel-E?

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

     Cheers,
     --MarkM



More information about the e-lang mailing list