Issue518

Title extra PDDL syntax checker
Priority feature Status resolved
Superseder Nosy List jendrik, kcleung, malte
Assigned To Keywords
Optional summary

Created on 2015-03-18.09:13:35 by kcleung, last changed by malte.

Messages
msg4099 (view) Author: malte Date: 2015-03-19.21:19:52
> With regards to the symbol table (issue 472), would you please let me have
> read access to the code?

Done! See https://bitbucket.org/malte/downward. (The branch is named issue472.)
msg4098 (view) Author: kcleung Date: 2015-03-19.20:35:08
That's great.  I am already more than half way in the implemented the syntax 
checkers for my boss.

With regards to the symbol table (issue 472), would you please let me have 
read access to the code?
msg4096 (view) Author: malte Date: 2015-03-19.16:51:29
> However, if I add the object type checking, and predicate checking features,
> would it be useful to contribute towards issue 215 and 220?

If the code is clean and correct, yes! :-)

I started working on some of these issues recently, but in many cases I hit a
roadblock because it's not really clear what the intended semantics is. PDDL is
unfortunately underspecified in some areas, especially w.r.t. the semantics of
types. But this shouldn't keep us from adding some of the obvious missing checks
in our code.

> Also in issue 472, it says that a symbol table will be introduced.  Does the
> current structure of the task object count as a symbol table?  Or will there
> be one in the future?

I've started working on issue472 -- if you want, I can give you read access to
that code on bitbucket. Maybe this answers the question. If not, let's continue
the discussion in the thread for issue472.
msg4093 (view) Author: kcleung Date: 2015-03-18.21:32:38
However, if I add the object type checking, and predicate checking features,  
would it be useful to contribute towards issue 215 and 220?

Also in issue 472, it says that a symbol table will be introduced.  Does the 
current structure of the task object count as a symbol table?  Or will there 
be one in the future?
msg4087 (view) Author: jendrik Date: 2015-03-18.10:38:13
Hi kcleung, it's nice that you want to contribute to Fast Downward. Please remember that only the core developers should 
change the fields "assigned-to" and "priority". Before we agree that we actually want to address an issue it should be 
assigned to nobody and its priority should be "wish" (unless of course it's a bug).

Regarding this particular issue: making the PDDL parser more rigorous is already at least partly adressed in issue215 and 
issue220. I am therefore marking this issue as resolved.
msg4086 (view) Author: kcleung Date: 2015-03-18.09:13:35
I plan to implement extra PDDL syntax checking, such as checking validity of the
types of the object in the object list, validity of predicates in the init, goal
and action lists etc.  I will work on a branch in my bitbucket repository and
let you guys have a look when I finish.

Do you find PDDL syntax checking is useful before translation to SAS?
History
Date User Action Args
2015-03-19 21:19:52maltesetmessages: + msg4099
2015-03-19 20:35:08kcleungsetmessages: + msg4098
2015-03-19 16:51:29maltesetmessages: + msg4096
2015-03-18 21:32:38kcleungsetmessages: + msg4093
2015-03-18 10:38:13jendriksetstatus: unread -> resolved
assignedto: kcleung ->
messages: + msg4087
nosy: + jendrik
2015-03-18 09:19:57kcleungsetnosy: + malte
assignedto: kcleung
2015-03-18 09:13:35kcleungcreate