search

GEGlobalResearch / DARPA-ASKE-TA1

ANSWER Project to demonstrate knowledge-driven extraction of scientific models from code and texts
Other
7 stars 5 forks source link

Content assist improvements #80

Open crapo opened 4 years ago

crapo commented 4 years ago

@kittaakos , another area of significance for improvement of user experience is content assist. You've worked on this in the past and the basic architecture is in place, but it doesn't work in a number of instances as expected. If you have any time available, let's work together to improve this. I can provide examples and what should be the completion options. Let me know if you would be available to spend some time on this.

kittaakos commented 4 years ago

If you have any time available,

I have, let's do it.

I can provide examples and what should be the completion options

Yes, please. The only way I can help if you provide the DSLs, the cursor positions and the expectations for the content assist, then I can implement those cases.

crapo commented 4 years ago

Here's a start. Suppose that I'm creating a brand new model, either in a .sadl file or a .dialog file. I have entered the following statements. Aircraft is a class. Engine is a class. engine describes With the cursor at the end of the last line, content assist gives the correct answers (Aircraft, Engine) along with the also correct answers of all of the classes in implicit models, all in a single alphabetized list. This list also includes all of the primitive data types, e.g., "double", but a primitive datatype cannot be a domain of any property so they should not be included here. (They can ony be ranges, so if the cursor were at the end of engine describes Aircraft with values of type then the primitive datatypes should be included. I'm wondering how hard it would be to a) use a SADL preference to include or not include implicit model concepts in content assist, and/or b) list the concepts in this model and any explicitly imported models in an alphabetized list followed by all of the implicit model concepts in a second alphabetized list.

kittaakos commented 4 years ago

Here's a start.

I can start with this, sure.

I'm wondering how hard it would be to a) use a SADL preference to include or not include implicit model concepts in content assist

This seems to be straightforward.

The things you have described aren't SADL specific? What does it do with Dialog?

crapo commented 4 years ago

these should behave the same way in both, so I assume it is implemented in SADL and inherited in Dialog.

kittaakos commented 4 years ago

so I assume it is implemented in SADL and inherited in Dialog.

Correct. Thanks for the clarification.

I am going to work on this starting tomorrow. If you have any other requirements, please add them to this issue.

crapo commented 4 years ago

I will add more.