[Termtools] LP and FP competition

[Peter Schneider-Kamp]

Dear all,

having the abstract query as a part of the input problem definitely
seems the way to go.

For compatibility with parts of the existing example sets and to avoid
that the abstract query produces a failure, I would propose the
following syntax (based on % being the Prolog character for line
comments):

%query: p(i,o).

This would specify all queries p(t_1,t_2) where t_1 is a ground
term and t_2 is arbitrary.

We could agree to have this comment line be the first line of
every file.

Some examples also use 'b' (for bound) instead of 'i' (for input)
and 'f' (for free) instead of 'o' (for output). There is also
a variant with 'g' (for ground) and 'a' (for any), but I don't
think there are any such examples in the TPDB.

What do you think?

Best regards,
Peter

Claude Marché wrote:
> Dear all,
> 
> First let me say that I like very much the new site for the termination 
> competition. I know this required a hard job so I'd like to congratulate 
> the developers for their efforts.
> 
> Regarding the queries for LP problems: indeed, each problem of the TPDB 
> contain some form of a query, but there was no consensus on the form of 
> queries. LP inputs are supposed to be correct prolog programs, but 
> queries, of the form P(i,o,i,..) are not valid clauses. The rules of the 
> competitions for the LP category explain that the query is given as an 
> extra argument. This extra argument was produced by the former 
> competition engine by recognizing each particular form of queries seen 
> in the TPDB files and making them under a common format.
> 
> Now, I think it would be better to decide of a common format of queries 
> directly inside the input file, so that this extra argument is not 
> needed anymore.
> 
> Hope this helps,
> 
> - Claude
> 
> 
> Johannes Waldmann wrote:
>>> 2) in the LP category, how do the tools determine the query mode? (is  
>>> it parsed from the input file or passed by the execution environment?)
>> it seems the mode is the first line of the input file,
>> see e.g. http://www.lri.fr/~marche/tpdb/tpdb-2.0/LP/talp/apt/SS_map.pl
>>
>> From looking at software used earlier
>> http://www.lri.fr/~marche/termination-competition/2006/talp/runme
>> it seems the query was given on the command line.
>>
>> I'd say, the participants of that category should state
>> what they'd find most convenient;  and then Simon checks
>> how it can be made to fit with the execution environment.
>>
>> (From a software engineering viewpoint, including the (mode and)
>> query with the problem statement would seem most reasonable:
>> then there is exactly one data source per problem execution.)
>>
>> Best regards, J.W.
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Termtools mailing list
>> Termtools at lists.lri.fr
>> http://lists.lri.fr/mailman/listinfo/termtools
> 



-- 
Peter Schneider-Kamp   mailto:psk at informatik.rwth-aachen.de
LuFG Informatik 2      http://verify.rwth-aachen.de/psk
RWTH Aachen            phone: +49 241 80-21211