Messages

1

Suggestions and Requests / See object representation in Intermediate language

« on: October 10, 2013, 12:17:47 am »

Hi there,

It would be a nice feature if you could select a model element and see the representation in intermediary language. With this feature in place one can model the intended product of an MDA transformation and see what the MDA transformation should produce. This will make authoring much easier.

Thomas

2

Suggestions and Requests / Re: File type and new programming languages

« on: January 26, 2011, 07:32:03 pm »

Works like a charm. Thanks!

3

Suggestions and Requests / File type and new programming languages

« on: January 25, 2011, 08:31:09 pm »

It should be possible to associate a default file type with a custom programming language.

I'm auto generating shell scripts using MDA transformations and code generating templates, but when I invoke the code generation the default file name is '<class name>.'

4

Suggestions and Requests / Dynamic diagrams linked to queries..

« on: December 21, 2010, 01:35:19 am »

It would be nice if the query mechanism in Enterprise Architect could be linked to diagrams in order to implement dynamic views of the model.

Example.
Configure a search for all elements with the phrase 'Controller' in the name. Link the search results to a class diagram to create a self-maintained diagram with all controllers in the application modeled.

5

Suggestions and Requests / Code engineering and version control

« on: December 16, 2010, 09:34:01 pm »

Add code generation and MDA transformation templates to version control.

In an entirely model driven project the templates constitute the code and these need to be part of the version control integration with the possibility to roll back, see differences etc. The generated model elements are (in my case) completely irrelevant since they can be erased and re-generated again (would be nice if that happened automatically).

6

Suggestions and Requests / Code enginering overview/debugging/ease of use

« on: December 10, 2010, 08:56:23 pm »

It would increase usability of the Code generation/MDA template framework tremendously if one or more of the following features where present:

1) Click on a model element to see exactly the set of code generation template set relevant for that class. Would be extremely useful to troubleshoot problems with code generation. For model elements generated by a MDA transformation in general it would be nice if a similar feature where present for MDA transformations. Click on a model element to see which MDA transformation templates where relevant for its creation.

2) Add a debug flag to the MDA/code generation framework which cause the insertion of traces to the generated model/code if set. For MDA transformations a tag could be added to a generated element bearing the template name responsible for its creation. For code generation comments (begin-end) could be inserted in the generated code when a template was started and stopped (including stereotyped overrides).

3) Make it possible to completely re-created MDA generated model elements simply by invoking a MDA transformation again. Right now every change to the MDA transformations, which removes/renames elements requires a manual deletion of elements generated previously followed by a new run of the MDA transformation. This is time consuming and error prone. In addition you have to specify the location of source code again which is also time consuming and error prone. Ideally the above was a one-click operation.

4) Organize the list of templates in a tree instead of a simple list to reflect the nested nature. There is already an excellent figure in the user manual.

7

Suggestions and Requests / Diagram double clicks and composite elements

« on: December 09, 2010, 08:57:40 pm »

Double clicking on a method/attribute on a non-composite class opens the properties window for the selected method/attribute.

For a composite class the embedded diagram is opened (even though a method/attribute has been selected). One would expect the properties window to be shown if a method/attribute has been selected.

This is extremely annoying since you don't get back to the original diagram once the embedded diagram is closed.

8

Suggestions and Requests / Re: Allow simultanous display of alias and name

« on: July 02, 2009, 04:03:51 pm »

Super - EA is a wonderfull tool!

9

Suggestions and Requests / Allow simultanous display of alias and name

« on: July 01, 2009, 08:51:35 pm »

It would be nice if it was possible to display both name and alias simultanously (e.g. <name>(<alias>).

Rationale.
We use auto counters to provide model elements with 'unique' numbers (a method for actually verifying this would also be nice).

When establishing traceability between design and test activities the name is used to identify the meaning of the model element whereas the alias (i.e. counter) is used to establish the identity of the model element.
In other words both are used at the same time.

10

Suggestions and Requests / Wish: Temporary "on-the-fly" diagrams

« on: December 23, 2008, 06:26:25 pm »

When maintaining a large model with many different perspectives it would be nice if I could select one or more elements in the model and have EA generate a temporary UML diagram with the selected element(s) and all other elements associated with the selected element(s). This would allow for quick sanity checks of the model.
Ideally it could be a 'hoover' feature (for single elements) as seen in the Cooliris preview plug-in for Firefox so I could traverse the model with the mouse and have these diagrams popping up as the element changes. Could also be included in seach results.

11

Suggestions and Requests / Wish: Dependancy between new element and document

« on: August 11, 2008, 06:36:08 pm »

Feature: When generating a new element from a document a dependancy should also be created from the the new element to the document element.

Rationale: Traceability.

Sample use: Currently we are doing stakeholder descriptions. We start out by describing the stake holders. Based on these descriptions we create actor elements. Right now we manually draw a dependancy from the actor to the document element.
Soon we will do the same thing for processes.

12

Suggestions and Requests / Re: Wish: Transformation refactoring/evolution

« on: January 23, 2008, 12:08:56 am »

I definitely want them to be connected.

But if I manually add a class C and at later time decide that this class should have been generated by my transformation, I would like to be able to add the creation of C to my transformation script and have the manually added class element interpreted as a result of the revised transformation.

In the special case where I've manually added two class elements with the same name to my model (in order to separate auto generated from manually entered code) I would like the revised transformation to hook onto the right class element.

13

Suggestions and Requests / Re: Wish: Transformation refactoring/evolution

« on: January 22, 2008, 11:35:56 pm »

An automatic feature would be nice and also refactoring through transformations.
However in some situations an automatic link mechanism would most likely cause problems (at least in my case). In my model I have multiple classes with the same name, which I use to separate auto generated code from the code made by hand (partial classes in C#). It is important that this situation can be handled.

14

Suggestions and Requests / Wish: Transformation refactoring/evolution

« on: January 22, 2008, 11:07:00 pm »

It would be nice if EA would allow the user to link the result of a transformation to existing model elements - i.e. a feature similar to the one found for code generation. This feature could be based on comparison of all attributes (name,  stereotype, tag etc) of the transformation result with the existing model and in case of a complete match the GUID from the transformation result should be assigned to the model element in question (user confirmation).

Rationale:
As the design of a system matures additional possibilities for generalization/standardization appear. Right now when these situations occur a new transformation has to be implemented and the transformation result (new elements) needs to be manually linked with the existing model elements as a replacement. This is costly and error prone and a barrier for changes to the transformations.

15

Suggestions and Requests / Wish: Code generation - bug fixing/tracebility

« on: November 22, 2007, 11:13:59 pm »

It would be nice if the names of the code generation templates used in a given file were somehow available in the file - either as a summary on a file/class/operation level or "near" the generated code. I'm doing this manually now but the structure of the information is scattered all over the source code, which makes the generated code "messy". It would be nice if the framework collected the summary based on the actual generated code and wrote a well-formatted section somewhere.

Rationale:
Implementing an extensive set of code generation templates with stereotyped overrides occasionally makes it difficult to track down the template used to generate a particular code fragment.