[abc] ecoop

Hi all,

I have finished a pass over the paper and corrected a number of typos.

I also noticed the following problems, which I will fix tomorrow unless
there is disagreement:

Section 4
"and how the aspects related to the Java IRs" => "relate", or "are
related"
 
We mention again that the grammar is a contribution. This was already
done in the into.

5.4
"This results in a substantial
amount of duplicated work, and makes the matcher unnecessarily
complex, in a straightforward implementation."
=>
"This results in a substantial
amount of duplicated work and makes the matcher unnecessarily
complex if implemented in a straightforward manner."

"One ambiguity in the existing specification of this feature is how it
behaves when combined with logical disjunction. is the treatment of
pointcuts which combine variable binding and disjunction."
(not sure about this one..what's that dot?)

6
"Furthermore Polyglot encourages
the use of many tiny classes, and it requires a fair amount of
boilerplate
for visitors and factories. "
Here we suddenly mention Polyglot. We need to remind readers that that's
our front end.

Regarding the compile time numbers:
Does anybody have a simple explanation why abc -O0 could be slower than
abc? I'll try to investigate tomorrow..

Cheers,

Sascha
Received on Sun Dec 19 22:46:06 2004