Issue413

Title Translator fails when object id and predicate id have the same name
Priority bug Status resolved
Superseder Nosy List gabi, sergionuc
Assigned To Keywords
Optional summary

Created on 2014-02-04.10:33:51 by sergionuc, last changed by gabi.

Files
File name Uploaded Type Edit Remove
thoughtful-typed.pddl sergionuc, 2014-02-04.10:33:51 application/octet-stream
thoughtful_11_6_53-typed.txt sergionuc, 2014-02-04.10:34:54 text/plain
Messages
msg2922 (view) Author: gabi Date: 2014-02-04.11:13:49
Sorry, I was too slow. Closing the issue again.
msg2921 (view) Author: gabi Date: 2014-02-04.11:12:59
Using the same id for objects and predicates is not allowed. Afaik, all newer
PDDL versions (>= 2.x) are extensions of the PDDL version used in the IPC 2000
and there it was explicitly forbidden:

"Names: the category <name> consists of strings of characters that beginning
with a letter and contain only letters, digits, hyphens (“-”),and underscores
(“_”). Case is not significant. <name>’s are required to be unique. That is, one
cannot use the same name in two different definitions."

Subset of PDDL for the AIPS2000 Planning Competition -- Draft 1.
Fahiem Bacchus.
Unpublished manuscript linked from the IPC-2 website, 2000.

However, it looks like this is actually not the problem in the attached task.
There is a predicate and a type of the same name (suit), but with different
arity and usage. This is not legal PDDL (for the same reason as above).

Allowing this, would also not be very nice because in the semantics it would
lead to a predicate symbol that it used with two different arities: "An atomic
type name is just a timeless unary predicate, and may be used wherever such a
predicate makes sense." (same reference as above).

Sergio, would renaming the type or the predicate solve the problem for you?
msg2919 (view) Author: malte Date: 2014-02-04.11:06:03
Hi Sergio, isn't this rather a conflict between a *type* name (rather than
object name) and predicate? The reason why the planner doesn't allow this is
because PDDL allows using types as predicates, but apparently this has fallen
out of fashion. I think this is a duplicate of issue293, so I'm closing this one
and adding a link to that one.
History
Date User Action Args
2014-02-04 11:24:04gabisetstatus: chatting -> resolved
2014-02-04 11:13:49gabisetstatus: unread -> chatting
messages: + msg2922
2014-02-04 11:12:59gabisetstatus: resolved -> unread
nosy: + gabi, - malte
messages: + msg2921
2014-02-04 11:07:12maltesetstatus: chatting -> resolved
2014-02-04 11:06:03maltesetstatus: unread -> chatting
nosy: + sergionuc
messages: + msg2919
2014-02-04 11:03:25maltesetnosy: + malte
2014-02-04 10:34:54sergionucsetpriority: urgent -> bug
files: + thoughtful_11_6_53-typed.txt
2014-02-04 10:33:51sergionuccreate