Re: [abc] oopsla paper

On Wed, 10 Aug 2005, Ganesh Sittampalam wrote:

> I'm going to have a go through it tonight (i.e. now).

I've finished my pass, though I wouldn't describe it as very complete. I
fixed a few small things.

One thing I noticed is that we did not seem to be very consistent in the
use of ':=' rather than '=' to indicate a definition. I decided that
trying to introduce ':=' in the right places was likely to be fairly
error-prone this close to the deadline, so I took the approach of just
removing them all. If there was any logic behind the ':='s that were
there, let me know and I'll revert this change.

>> 2) In Related Work, near the end of the section on Douence et al, is
>> the phrase "in that paper, the code is LITTERED with explicit equality
>> tests". Were we supposed to be nice to them in the related work? Is
>> "littered" the word we want to use?
>
> I agree that we should change this. We can give the same general impression
> without sounding quite so insulting, e.g. "the code has many explicit
> equality tests".

I've now changed this. Feel free to change it further, of course.

My diff is below.

Ganesh

Index: Text/intro.tex
===================================================================
RCS file: /srv/cvs/oopsla05/Text/intro.tex,v
retrieving revision 1.11
diff -u -u -r1.11 intro.tex
--- Text/intro.tex 10 Aug 2005 14:45:34 -0000 1.11
+++ Text/intro.tex 10 Aug 2005 23:38:15 -0000
@@ -41,7 +41,7 @@

  Although our examples provide a general feeling for our new tracematch
  language feature, we felt that it was important to give a rigorous
-definition and use this definition to lead to a correct and sound
+definition and to use this definition to produce a correct and sound
  implementation. We first define a reasonably obvious
  declarative semantics, then we give a non-trivial operational semantics
  that could be used to guide a reference implementation, and finally we
Index: Text/related.tex
===================================================================
RCS file: /srv/cvs/oopsla05/Text/related.tex,v
retrieving revision 1.15
diff -u -u -r1.15 related.tex
--- Text/related.tex 10 Aug 2005 14:45:34 -0000 1.15
+++ Text/related.tex 10 Aug 2005 23:44:18 -0000
@@ -87,7 +87,7 @@
  of the former are discussed in \cite{cibran05} and of the latter in
  \cite{fritz05}. Especially the examples of \cite{fritz05} provide strong
  indication of the importance of matching with variables as we
-have defined it: in that paper, the code is littered with explicit
+have defined it: in that paper, the code has many explicit
  equality tests between variables. In tracematches, such equality tests
  are expressed by simply using the same variable multiple times.

Index: Text/semdecl.tex
===================================================================
RCS file: /srv/cvs/oopsla05/Text/semdecl.tex,v
retrieving revision 1.16
diff -u -u -r1.16 semdecl.tex
--- Text/semdecl.tex 10 Aug 2005 14:45:34 -0000 1.16
+++ Text/semdecl.tex 10 Aug 2005 23:53:51 -0000
@@ -147,7 +147,7 @@
  The event set of a tracematch $tm$ under a given valuation $\sigma$ is
defined to be the
  set $\Omega(tm, \sigma)$ of events that are matched by some defined
symbol in $tm$, with
  variable bindings compatible with the valuation $\sigma$. Formally, we
define:
-\[\Omega(tm, \sigma) := \{e \in \textit{event} \mid (\exists a : a \in A
: \sigma(a(e)) = \textit{true})\}\]
+\[\Omega(tm, \sigma) = \{e \in \textit{event} \mid (\exists a : a \in A :
\sigma(a(e)) = \textit{true})\}\]

  Finally, we write the trace obtained from $t$ by removing any events not
in a set $S$ as $\filter{t}{S}$.

Index: Text/semop.tex
===================================================================
RCS file: /srv/cvs/oopsla05/Text/semop.tex,v
retrieving revision 1.7
diff -u -u -r1.7 semop.tex
--- Text/semop.tex 16 Mar 2005 15:06:43 -0000 1.7
+++ Text/semop.tex 10 Aug 2005 23:42:13 -0000
@@ -12,7 +12,7 @@
  of inconsistent variable bindings. This is defined by:
  \[
  \begin{array}[t]{l}
- \SKIP(e) := {}\\
+ \SKIP(e) = {}\\
     \lnot (\lor a : a \in A : a(e)) = {}\\
     \land a : a \in A : \lnot a(e)
  \end{array}
Index: Text/semtoimpl.tex
===================================================================
RCS file: /srv/cvs/oopsla05/Text/semtoimpl.tex,v
retrieving revision 1.9
diff -u -u -r1.9 semtoimpl.tex
--- Text/semtoimpl.tex 10 Aug 2005 14:45:34 -0000 1.9
+++ Text/semtoimpl.tex 10 Aug 2005 23:49:39 -0000
@@ -10,7 +10,7 @@
  During execution, each state $s$ of $M$ is labelled by the
  constraint
  \[\LAB(s, t) = \textit{match}(L(s),t)\]
-, where $t$ is the current trace.
+where $t$ is the current trace.
  It is shown below how to update these constraints when a
  new event is appended to the current trace. After we have
  computed the
@@ -31,7 +31,7 @@
  %We now assume that $s$ is not initial in $M$ in the following.

  We define $\LAB(s, t)$ by recursion on $t$. The base case is
-\[\LAB(s, \varepsilon) := \textit{match}(L(s), \epsilon) =
+\[\LAB(s, \varepsilon) = \textit{match}(L(s), \epsilon) =
  \left\{
  \begin{array}{ll}
  \textit{true} & \textrm{if $s$ is an initial state}\\
Received on Thu Aug 11 00:56:48 2005