Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Uffe

Pages: 1 ... 52 53 [54] 55 56 ... 85
Suggestions and Requests / New Baseline: Set Package Version
« on: April 09, 2010, 06:36:37 pm »

When I create a new Baseline, I would really really REALLY like the option of automatically setting the package's Version to whatever version identity I select for the new baseline.



Suggestions and Requests / Re: Diagram: Find in Project Browser
« on: April 09, 2010, 06:33:18 pm »
Boooooy, is my face red.  :-[

Suggestions and Requests / Diagram: Find in Project Browser
« on: April 09, 2010, 05:54:09 pm »

I can select an element in a diagram and via its context menu (or by pressing Alt-G) Find it in the Project Browser.
I'd like to be able to do that with the diagram itself, through its context menu or by pressing Alt-G when nothing in the diagram is selected.
Helpful when modelling in several different branches of a large model. Plus it'd take you guys like five minutes to implement. :)



Suggestions and Requests / Feature Visibility: Version
« on: March 29, 2010, 07:01:19 pm »

I'd really like to see the option of switching on a "Version" compartment for elements in diagrams. Or perhaps even "Details," which would show version and author (similar to Diagram Details).



Suggestions and Requests / GUI Feature: Select Connected
« on: March 18, 2010, 08:37:43 pm »

Here's a GUI feature I'd like: select one or more elements in a diagram, right-click, choose "Select Connected."
This causes the GUI to add to the current selection all (previously non-selected) elements in the diagram which are connected to one of the elements in the original selection.
Great for moving things around in a complex diagram with hand-drawn layout.


Suggestions and Requests / Code Template Framework: add @sort to %list%
« on: December 09, 2009, 07:44:01 pm »

The code generation control macro %list% could use a @sort argument. Things often need to appear in a resulting file in a strict order, defined by the contents of the elements.

The argument should be the name of a macro to be applied in context when the %list% is executed. Examples:
%list="Class" @sort="className"%
%list="Operation" @sort="opAlias"%
%list="Attribute" @sort="attInitial"%
%list="Namespace" @sort="packageTag:%qt%MyTag%qt%"%

It should also be possible to specify a reverse sort by adding an additional argument @reverse="True".

Not specifying a @sort argument results in the model default order (which can still be reversed).

This functionality would be immensely useful for some of the stuff we do over here.



Suggestions and Requests / Model Transform: Optional Diagram Creation
« on: November 18, 2009, 07:02:12 pm »

I work with MDA Transforms a lot, and while the auto-generated diagrams are often useful, other times they are not.

Since diagram creation and layout accounts for a significant proportion of resource usage during a transform, I'd like a "Create Logical Diagrams" tickbox in the Model Transformation dialog.



Suggestions and Requests / Element Properties: Link End
« on: August 19, 2009, 10:16:40 pm »

The Link tab of the Properties dialog could do with an indication of which end of the link the element is at, source or target.



The Code Template Framework allows access to Class constraints only, but Attributes, Connectors and Packages can all have constraints too.

I suggest the following:
  • Rename the macros constraint* -> classConstraint*
  • Add attConstraint*
  • Add connectorConstraint*
  • Add packageConstraint*
  • Remove connectorSourceConstraint and connectorDestConstraint (connectorSourceElem* and connectorDestElem* should take care of those)
To me personally, the attConstraint* ones are the highest priority.



Suggestions and Requests / Re: Package: Set As Namespace Leaf
« on: August 11, 2009, 04:25:59 pm »
EA does indeed not enforce unique naming of elements within a package which means, by induction, that you can end up with multiple elements with the same fully-qualified name.

I've written up a separate feature request to help with this issue, which I believe goes right back to the UML standard:

The "namespace leaf" feature was not intended to address the uniqueness issue, but to allow display of (two) different source packages for one element depending on display context, purely for readability purposes when using recurring substructures in a model.


Suggestions and Requests / Package: Set As Namespace Leaf
« on: August 06, 2009, 06:08:23 pm »

This feature is intended to improve the display of foreign elements (ie elements from a different package) in a diagram, primarily within large models with recurring package structures.

Let's say you have a model along these lines:
+-- Software Modules
|   +-- Blarg (*)
|   |   +-- Actors
|   |   +-- Use Cases
|   +-- Honk (*)
|   |   +-- Actors
|   |   +-- Use Cases

Ignore the (*) for now.
Let's say in one of my Honk use cases I wish to include a use case from Blarg.
When presented in Honk's diagram, this use case will be labelled "(From Use Cases)".
Since I am using a recurring package structure, this label isn't very helpful: EA displays only the immediate parent package of the foreign element, and that package name occurs in dozens or hundreds of different software modules.

I propose adding an option "Set As Namespace Leaf" to a package. In the structure above, assume I have applied this flag to the (*) packages.

When displaying a use case symbol from Blarg - Use Cases in a diagram in Honk - Use Cases, EA would then label it "(From Blarg)".
In other words, EA searches upwards in the package hierarchy and uses the name of the first package it finds with the "Namespace Leaf" flag set.
If the search hits the root node with no "Namespace Leaf" package found, EA uses the name of the element's immediate parent package (ie today's behaviour).
In a use case diagram in Honk - Use Cases, an actor from Honk - Use Cases should be labelled "(From Actors)".
This is achieved by EA performing the "Namespace Leaf" package search twice, once from the element and once from the diagram.
If the first hit in both searches is the same package, ie the element and diagram have the same nearest "Namespace Leaf" package, the immediate parent package name is displayed (again, as today).

This way, Honk actors will be labelled "(From Actors)" in the Honk use case diagrams.
In the same Honk use case diagrams, use cases from Blarg - Use Cases will be labelled "(From Blarg)".
Let's say that one of my Honk actors generalizes a Blarg actor. In the Honk actor diagram the Blarg actor will be labelled "(From Blarg"), while in a Blarg use case diagram, the same actor will be labelled "(From Actors)".

The package symbol in the project browser should be different when the package is a "Namespace Leaf."
Finally, "Namespace Root" is a code engineering concept, while "Namespace Leaf" is a purely visual one. Therefore the option shouldn't go in the same place as "Set As Namespace Root," and perhaps the name "Namespace Leaf" is a bad one.



Suggestions and Requests / Unique Element Names within Package
« on: August 11, 2009, 04:20:47 pm »
Hey everyone,

This feature takes off from a discussion regarding namespaces and uniqueness of element names that came up in

EA does not enforce unique element names within a package, and I'm pretty sure it cannot do so and remain UML compliant. The standard only says that an element must have a name, it doesn't say anything about uniqueness.

However, names within a context do need to be unique in order for the model to be properly understood - either by a person or by a compiler.

So I propose adding an Object option, "Warn about duplicate element names", which would flash a confirmation dialog whenever you tried to give an element a name that another element in the same context already has, ie element in package, class in outer class, etc.
This would apply to name entry during create/rename, as well as when moving an element from one context to another. You wouldn't want it during imports, although perhaps it might be useful to see a list of such offending elements in the import output.

In addition, a model validation rule should be added to check for uniqueness of element names within each context.



Seconded! (Or thirded, or fourthed, or whatever.)

Locks do sometimes need to be managed manually and the current dialog does not let you do that very effectively, either as a regular user (manage own locks) or as a "security manager" (manage all locks).

The current dialog lets you select multiple elements (or all) and "unlock selected".
A tree view would be more useful here, along with the options "unlock selected only" and "unlock recursive".

And for full lock management functionality, add the display switches "unlocked only", "locked only" and "locked and unlocked", plus buttons to lock the selected elements.
This, I stress, is nowhere near as important as getting a more useable unlock dialog.

Suggestions and Requests / Select Type Dialog: Language Types
« on: August 03, 2009, 08:25:31 pm »

Loving the Select Type dialog in 7.5.846.

Would like to suggest the following addition: add a root node to the tree for the element's specified Language, which contains the Language's built-in types. So you'd see something like:

+ My Model Root Node 1
|  +--- My Model Root Node 1 View 1A
+ My Model Root Node 2
|  +--- My Model Root Node 2 View 2A
|  +--- My Model Root Node 2 View 2B
+ Built-In Java Types
|  +--- boolean
|  +--- byte
|  +--- char


The language-specific node would have to be context-sensitive of course, just like the quick type selector drop-down in the Attribute Properties dialog, and you wouldn't be allowed to Add New in this branch.

This would also help fix a bug in the current implementation: if your attribute is of a built-in type and you open up the Select Type dialog, a (non-builtin) element is highlighted in the tree, indicating the wrong type for the attribute.



Suggestions and Requests / Subversion Credentials
« on: July 15, 2009, 03:51:19 am »

When working with Subversion, EA assumes that the user logs into the repository using some other mechanism, and hangs during check-out if this hasn't been done.

The hang is obviously a bug, but the feature I'd like added is a way to enter your Subversion credentials for EA to supply to the server.


Pages: 1 ... 52 53 [54] 55 56 ... 85