| Issue with using Haetae [message #1782150] |
Mon, 19 February 2018 14:48  |
 Djamel Khelladi
Messages: 1
Registered: February 2018 |
Junior Member |
|
|
Hi,
I am trying to use Haetae because I have been told that it can provide type information for elements in an ETL transformation.
I have a first question. I installed Haetae through the update site and then I ticked "enable static analysis (experimental)". in Preferences > Epsilon and Epsilon haetae, before I registered the two metamodels source and target used in an ETL transformation.
The first Error was that I could not transform an ETL with haetae because the method getAST in EolLibraryModule was not found (in my 1.4 version of epsilon).
So I installed the 1.5 interim version of Epsilon.
Now I do not get any error, but rather a dialog saying that "the chosen operation is currently not available" when I try to use the 3 choices in the menu of Haetae.
I also tried the run the available Haetae code on github, but it does not work, becasue now in version 1.5 of epsilon, the class EolLibraryModule is not there anymore (either was deleted or moved).
Please, can you help me here?
Another question, does Haetae provide information on the container of elements in addition of their types.
For example, if I have an ETL assignment"target.date.startDate = ...."
Can Haetae gives me the following information:
1) the element target is of type class C contained in package P
2) the element date is of type Date contained in class C
3) the element startDate is of type string contained in class Date
Hope I detailed enough my issue :)
Best regards.
|
|
|
| Re: Issue with using Haetae [message #1782252 is a reply to message #1782150] |
Tue, 20 February 2018 22:18  |
|
Hi there,
Haetae has not been maintained to adapt to Epsilon since v1.3, so falling back to v1.3 may help configure Haetae correctly. You are correct in the case that Epsilon structure has been changed quite significantly.
With regard to your second question, Haetae provides a "best guess" type resolution for expressions, so it will give you better results when you provide actual type information of variables (in this case the target of the transformation, which should be declared in your rule signature, so that you can get the type infomation on 2) and 3)).
I hope this helps.
Research Associate
Department of Computer Science
University of York
|
|
|
Powered by
FUDForum. Page generated in 0.02954 seconds
] 
Search
Help
Register
Login
Home
