Project

General

Profile

Bug #1543

create workflow scripting language

Added by Chad Berkley over 15 years ago. Updated about 15 years ago.

Status:
Resolved
Priority:
Low
Assignee:
Category:
interface
Target version:
Start date:
04/30/2004
Due date:
% Done:

0%

Estimated time:
Bugzilla-Id:
1543

Description

it would be useful to have a simple scripting language to write out workflows
without using the visual editor and without having to write a moml file by hand.
this could be done using javacc.

History

#1 Updated by Christopher Brooks over 15 years ago

Ptolemy II includes Jacl, which is a 100% Java implementation of a
subset of Tcl. We use Jacl for the test suite. Tcl is fairly easy
to pick up, and there are plenty of examples about how to run models.
There is some documentation about Jacl at
http://ptolemy.eecs.berkeley.edu/ptolemyII/ptII4.0/ptII4.0-alpha/doc/coding/testing.htm

Another language to look at is Jython, which is a Java implementation
of a subset of Python. There is also a Python script actor that
might be useful. See
http://www.jython.org/

I think Python probably has a larger user community than Tcl these days.
Both Jacl and Jython are not really being actively developed, which is
good in that they are not wildly changing over time. I've been fixing
bugs in the Ptolemy copy of Jacl, and there is als a Sourceforge TclBlend/Jacl
project. Last I looked, Jython had not been updated in a couple
of years, but we recently recompiled Jython to solve a Java 1.5 runtime
problem.

Another language to look at is Caltrop, which is a language that also shipped
with Ptolemy II:
"Its research focuses on the definition of CAL, a textual notation for writing
dataflow actors, and uses and applications of that language inside and outside
of Ptolemy and the Ptolemy II software."

Ptolemy II uses JavaCC in several places, but you might want to look
at using CUP and JFlex instead of JavaCC. Basically, using JavaCC
as a LL parser can result in ambiguities as a language grows. I'm
told, and I vaguely remember, that LALR parsers such as CUP/JFlex tend
to have fewer problems with ambiguity.

One trick we use when defining models in Java is to add an exportMoML()
call.

For example, ptolemy/domains/sdf/demo/OrthogonalCom/OrthogonalCom.java
says:
// Uncomment the next line dump out xml.
// System.out.println(exportMoML());
One can do similar things in Jacl and presumably Jython as well.

Basically, I encourage you to look at an off the shelf language and see
if you can use it. One trick is to define some utility procedures to
do common tasks. For example, in Ptolemy II, we have Tcl code in
$PTII/util/testsuite that defines sdf and de models. These utility
methods are used to set up models for testing.

-Christopher

#2 Updated by Chad Berkley about 15 years ago

Since this is already implemented via JACL in PTII, this bug is moot.

#3 Updated by Redmine Admin over 6 years ago

Original Bugzilla ID was 1543

Also available in: Atom PDF