|
Re: EOL [message #1016800 is a reply to message #1016767] |
Thu, 07 March 2013 17:52 |
|
I'm sorry, but the E*L editors do not support autocomplete. The Epsilon languages support models which do not have metamodels, so we can't rely on having something to autocomplete against.
|
|
|
|
Re: EOL [message #1829754 is a reply to message #1829751] |
Fri, 10 July 2020 08:37 |
|
Hi Thomas,
Yes, this is still accurate. We are working on static analysis for Epsilon programs but our first milestone is to use it to provide metamodel-aware error checking. Code completion would be the next logical step but we don't have a timeline for this. To get a preview of the (currently very incomplete) static analysis facilities, you can turn on static analysis in the Eclipse preferences dialog (Epsilon) and then try the following code in an EOL file:
model M driver EMF {nsuri="http://www.eclipse.org/emf/2002/Ecore"};
var c : EClass;
var u : Unknown;
Thanks,
Dimitris
|
|
|
|
|
Re: EOL [message #1835009 is a reply to message #1834942] |
Mon, 23 November 2020 09:59 |
|
Hi,
Thanks for offering to help! We are currently working on static analysis facilities for EOL [1] that allow us to infer the type of EOL expressions, perform type checking and report relevant errors (e.g. when attempting to assign a string value to an integer property) in Epsilon editors.
What we haven't looked into so far is the mechanics of offering context-aware code completion within the Epsilon editors. If you could develop a small proof of concept that extends the EOL editor to e.g. propose a few fixed methods every time '.' is pressed after a string literal, that would certainly help speed things up.
Thanks,
Dimitris
[1] https://github.com/quratulain-york/org.eclipse.epsilon
|
|
|
Re: EOL [message #1841609 is a reply to message #1835009] |
Sat, 22 May 2021 07:06 |
jo ber Messages: 94 Registered: August 2018 |
Member |
|
|
Hi
I thought about an alternative approach:
1. Convert the Epsilon grammar to Xtext grammar
2. Use the generated code in xtext for validation and autocompletion
Do you see any flaws in this approach?
My main issue would be to convert the grammar. I'm not really experienced in this area.
Kind regards
[Updated on: Sat, 22 May 2021 07:08] Report message to a moderator
|
|
|
Re: EOL [message #1841610 is a reply to message #1841609] |
Sat, 22 May 2021 07:15 |
|
Hi Jo,
This would be no small feat I'm afraid. Beyond converting the EOL grammar, you would also need to replicate the logic that resolves type names to metamodel types (in the different modelling technologies supported by Epsilon), property call expressions to attributes/references, operation calls to built-in/user-defined operations etc. Moreover, you'd need to keep all this in sync as Epsilon evolves over time.
Best,
Dimitris
|
|
|
Powered by
FUDForum. Page generated in 0.03943 seconds