Comments on the format of the termination problems of the TPDB

[Aart Middeldorp]

Dear all,

Thanks for Albert and Claude for the first draft and
Alfons for initiating the discussion! To simplify things,
I will not quote any messages, but try to make some general
remarks.

(1) The purpose of the format is not that it is easy to
     read by humans, but that it is easy to manipulate by
     computers and that it is extendible to cover future
     situations (like higher-order).

(2) Some of the discussions on the syntax are not really
     important; rather we can leave out a lot of the syntax
     decisions if we adopt XML.

For instance, it doesn't matter whether we write "->" between
the two sides of a rewrite rule, or whether we separate two
rewrite rules by a linefeed, a semi-colon, or whatever. To
give a concrete example, the XML code for the two-rule system

f(a) -> b
    b -> c

could contain

<rules>
<rule>
<lhs>
f(a)
</lhs>
<rhs>
b
</rhs>
</rule>
<rule>
<lhs>
b
</lhs>
<rhs>
c
</rhs>
</rule>
</rules>

Concerning conditional rules, again XML avoids the choice
between '|' (which is symmetric so confusing) or some scheme
with parentheses. For example, the rule

a -> b <= c -><- d, e ->* f

could be written as

<rule>
<lhs>
a
</lhs>
<rhs>
b
</rhs>
<conditions>
<joinability>
<lhs>
c
</lhs>
<rhs>
d
</rhs>
</joinability>
<reachability>
<lhs>
e
</lhs>
<rhs>
f
</rhs>
</reachability>
</conditions>
</rule>

Another advantage of XML is that we don't have to worry
about the suffix naming conventions; ".xml" suffices.

Concerning the syntax of function symbols, surely we can
agree upon a regular expression that includes the common
mathematical symbols but forbids expressions like "&%(*&_,*^)".

-- Aart