[E-Lang] Progress & Non-Progress Report

[Mark S. Miller]

                               Sorry, I ran out of time

Not since 0.8.9 (and its recent offshoot, 0.8.9.1) has a release of E 
included a working CatTP layer.  In the meantime, the language itself, as 
represented by the internal non-distributed 0.8.9t release, has evolved 
significantly since 0.8.9[.1].  E developers are currently faced with the 
unpleasant choice between programming in a distributed E before various 
syntactic reforms, and programing in a modern version of the language that 
isn't yet distributed.

I've been working to repair that situation in a timely manner.  Besides the 
usual time pressure, I'm deeply honored to be invited to speak at "Austrian 
Perspectives on the New Economy" 
http://panoramix.univ-paris1.fr/AHTEA/colloques.html in Italy.  I'll be 
leaving for this on Monday May 21.  I had desperately wanted to get this 
next release out before I left, but I just ran out of time.  Between now and 
when I leave I'll be preparing my talk and preparing for the trip.  Terry 
and I will then take a needed vacation. We'll be back on June 10th or so.

My deep apologies to all using E.  I don't expect to be making further 
changes to the source tree between now and when I leave, except small urgent 
fixes as needed to support current users.  If you need something, be sure to 
tell me soon, or wait till June 10th, or be prepared to make the fix yourself.

Current E users: If you want the option of contacting me for urgent help 
while I'm travelling, contact me privately before I leave and I'll tell you 
how to get me.

                                The Lost Resolution Bug

The main reason this all took longer than expected is that I ran into a 
conflict between a set of desired attributes of E, now documented at 
http://www.erights.org/elib/distrib/captp/WormholeOp.html and 
http://www.erights.org/elib/equality/same-ref.html#lost-resolution .  The 
unimplemented interim solution documented on those pages owes a huge debt to 
many hours-long conversations with Dean Tribble and Marc Stiegler.  Thanks!

When we first realized the conflict, we hadn't realized that "Allow services 
with Near arguments" (see the WormholeOp page) should be a requirement.  And 
elevating it to a requirement contradicts the position I took in the 
previous correspondence with Tyler on e-lang regarding partial order.  
Nevertheless, as we discussed it, the consequences of backing down from this 
property was an inability to write simple services simply, and that would 
not have been acceptable.  (I believe this realization corroborates some of 
Tyler's arguments in that correspondence.)

I've already discussed with Bill the changes needed to VatTP to support 
WormholeOp, and he's prepared to do them.  In the meantime I'm planning to 
implement 0.8.10 so that it suffers from the Lost Resolution bug (as 
documented at the 2nd URL) in order to postpone actually doing the 
WormholeOp. Therefore, there's time to examine these issues before we start 
investing too heavily in any one solution.


        Cheers,
        --MarkM