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.

Topics - Uffe

Pages: [1] 2 3 ... 23
Bugs and Issues / ShapeScript NotesVisible property
« on: February 28, 2020, 06:48:20 pm »
Hello everybody,

In 1310 the NotesVisible property was introduced in shape scripts:
  • NotesVisible - Available for any element, allows querying if user has requested display of notes for this object or this diagram

Can't get it to work in 1512.

HasProperty("NotesVisible") evaluates to true, so the property does exist. But how do I set it?

I've tried both the diagram's properties (Elements - Notes) and the element's compartment visibility. Either way, the NotesVisible property is stubbornly stuck at false.

What am I doing wrong?


Bugs and Issues / Resize makes element ginormous in diagram
« on: February 28, 2020, 06:21:38 pm »
Hi all,

I'm on 15.0.1512, and there's a bug in the element resizing. Sometimes when you grab a corner and start dragging, the element becomes absolutely huge.

If you hit escape before releasing the mouse button, the element reverts back to its previous size.
If you release the mouse button, the element gets the new ridiculous size.
You can undo the change, but the behaviour persists for that element until the diagram is closed and reloaded.

I've noticed this in activity diagrams, but it may not be limited to those. Either way, this is a gigantic productivity killer.

In the history for 15.0, there is one possibly relevant entry
Quote from: 15.0.1513
  • Objects will no longer resize the top-left edges unexpectedly when a base zoom is applied

Is this what that refers to? Because I'm observing this behavior without making any changes to the zoom.
Or does it perhaps mean that the bug strikes when the diagram is in the 100% zoom maybe? In which case it does resolve this?



Bugs and Issues / Send / Receive in QuickLinker
« on: February 28, 2020, 06:10:35 pm »
Hi all,

I'm on 15.0.1512, and I'm having issues with Send and Receive elements and the QuickLinker. I've got some stereotyped Actions (Atomic and CallBehavior), and I'm using unstereotyped Send and Receive (the red ones).

The problem is that using the QuickLinker in my custom diagram I don't get a ControlFlow option in the popup menu when drawing a connector from either a Send or Receive to one of my actions. Furthermore, I can't get one to appear by specifying it in my profile's QuickLink artifact either.

Yetfurthermore, if I create a vanilla activity diagram using unstereotyped everything, ControlFlow also does not appear in the popup menu.

Control flows are clearly allowed going from a Send or Receive to an action. Not only logically and by the standard, but also by the manual -- look in the sample diagrams in the manual pages linked above.

So what's going on here?

Why can't I draw a ControlFlow from a Send or Receive to an Action?
Why can't I force it with a QuickLinker definition?

And how do I work around this bug?



Bugs and Issues / Wrong diagram author shown in Properties window
« on: February 21, 2020, 06:22:47 pm »
Hi all,

On 15.0.1512 and a SQL Server repository, the Properties window shows the logged-in user's name in the Author field when t_diagram.Author is null. The window behaves correctly with a null-Author element/t_object.
In this repository we've also got t_authors set up, so that may be a contributing factor.

Can someone check if this has been fixed in a later version?
There's nothing in the histories.


Hi guys,

When you generate a document through the GUI you can include a watermark.
Is there a way to include one when you're using the DocumentGenerator?



Automation Interface, Add-Ins and Tools / Linked document templates
« on: February 19, 2020, 03:56:45 am »
Hi all,

Can linked document templates be packaged into an MDG Technology?



Suggestions and Requests / Decode proprietary content in tooltips
« on: February 17, 2020, 11:16:51 pm »
Hi all,

If you hover over an attribute in the properties dialog, properties window or features window, EA flashes a tooltip with the attribute's initial value.
If you're working on a profile with shape scripts, and hover over the _image attribute, the result is a huge tooltip full of gibberish.

It would be useful if EA decoded the shape script and showed that instead of the encoded data.

Possibly there are other situations as well where EA shows raw proprietary data in a tooltip, which it could decode for a better user experience.


Bugs and Issues / Multiple same-name tagged values in stereotype
« on: February 17, 2020, 09:07:45 pm »
Hi metamodellers,

I'm working on a profile with a stereotype where I want a tagged value to have multiplicity [1..*].
So I create an attribute for the tag in the stereotype, fine, works.

But when I use the resulting profile and want to add additional tags with the same name, things break down.

Let's say my profile is name TheProfile, my stereotype TheStereotype and my tag TheTag.

If I create an element of the relevant stereotype, EA presents TheTag
  • In the properties dialog, in a tab TheProfile (and not in the Tags tab)
  • In the properties window, in an expansion node TheStereotype in the Element tab (and not in the Tags tab).
If I create a new tagged value and call it TheTag, EA presents that
  • In the properties dialog, in the Tags tab (and not in the TheProfile tab)
  • In the properties window, in the Tags tab (and not in the Element tab).
That's not what I want. I want it alongside the other tagged values defined in the profile. This is just another instance of one of those.
So I now create a new tagged value and call it TheProfile::TheTag. EA presents that
  • In the properties dialog, as TheProfile::TheTag in its own expansion node in the TheProfile tab (and not in the Tags tab)
  • In the properties window, not at all.
This also is not what I want.

I'm on 15.0.1512, and according to the history, in 1514 there was a UI change to "Display non profile tagged values that contain '::' in their name". No further changes flagged in 15.1.

So is it now possible to do what I want? That is, add additional instances of a tagged value defined in a stereotype and have them displayed alongside their neighbours, just like EA does with non-stereotype-defined tagged values?


Bugs and Issues / Prohibited stereotype tagged value names
« on: February 17, 2020, 08:32:33 pm »
Hi everyone,

I'm working on a profile with a stereotyped Activity.
The Activity is supposed to have a tagged value "method" with an enum value.

I can model this in the usual way, but when EA creates the profile it replaces the name "method" with "Referenced Behavior".
The set of enum values is correct, though.

I can't find any property of a non-stereotyped Activity named "method", and there isn't one in the metaclass attributes either.
And I haven't checked whether this is exclusive to Activity stereotypes, or more general.

Is anyone aware of any other tagged value names EA has problems with?
I'm on 15.0.1512.


Hello hdocument hackers,

I'm working with the DocumentGenerator and I want to create a table where a couple of the columns have sub-rows -- or put another way, some of the cells in a row are split vertically.

Let's say there are two columns in my main table. The first column is to contain element names, and the second the names of the connectors going from the element. These connector names should be in separate cells (what I call sub-rows), but looking at the main table, the whole set of connectors should be in the same row of as the element.

My question is: is there a way to construct
a) a document template, and
b) an XML tree that EA will understand?

Normally, the data passed to DocumentGenerator.DocumentCustomData contains a number of <Row>s, each with a number of custom-named elements which EA matches to the custom fields in the template. But is it possible for one of those XML elements to be another <Row>?
Code: [Select]

Sort of thing?



Hi all,

I've got a shape script for a CallBehaviorAction, which looks similar to an EA "extended" process, which I want to decorate with the alias ("partition") in the NW or SW corner. But since the shape is a process, putting text there means one of the contour lines will run through it.

So I decided to add a couple of blank spaces before the text: Print("  #Alias#");. This doesn't work.

With one or two blank spaces, EA ignores them.

With three or more, EA moves the text down one line (and otherwise ignores the number of blanks).

It's not a wrapping thing. If instead of the blank spaces I use full stops, I can put in dozens of them and EA prints the whole thing happily without any line feeds.

I tried it with a '\t' instead of the blanks. EA interprets this as a tab... but then does the same thing: moves the text to the beginning of the next line.

How can I move the text slightly to the right?


Automation Interface, Add-Ins and Tools / Hyperlinks in document generation
« on: February 13, 2020, 03:28:21 am »
Hi all,

I've got a DocumentGenerator script, which incorporates an element's LinkedDocument, if any, like so:

Code: [Select]
if element.GetLinkedDocument() <> "" then
docGen.InsertLinkedDocument element.ElementGUID
end if

This works in that the linked document is indeed included in the generated document. However, EA buggers up any hyperlinks in the linked document but royally.

All hyperlinks work correctly in the linked document, first of all. I can follow them from there in EA's RTF editor, it fires up the default browser which displays the pages happily.

Bug #1: if the display text of a link contains non-English characters, EA creates multiple hyperlinks in the generated document.

So a link that displays the text Fáncie and that points to, will in the generated document have turned into 3 (three) hyperlinks with the display texts F, á, and ncie, each pointing to the same URL. Since they are displayed right next to one another in Word, it looks like one and the same hyperlink, but if you edit one of them you'll see the others don't change.

Bug #2: hyperlinks in EA's internal format break completely. This is because EA doesn't strip out the $inet: from the URL when generating the document. Thus, if you create a hyperlink in the document using right-click -- Create -- Hyperlink -- Web Site (in the dialog), it gets mangled. Hyperlinks created using Document Edit -- Insert -- Hyperlink work.

The mangled links also become file links, but I'm pretty sure that's because Word assumes a malformed URL must be an internal in-Word link, or at least a file link, and tries to do the best of a bad job. Which doesn't work.

Bug #3: fragment URLs get lost completely. EA splits the URL at the '#', places the fragment identifier into the hyperlink but never puts the first part back -- the one with the protocol, server and resource. Why it does this I can't imagine. While the fragment reference is useful, the important part is actually the first bit. I can always scroll the page if I need to, but guessing which one of the X billion pages out there my fragment is at is a little harder.

Using the Word API on the resulting file, the Hyperlink object has an empty .Address, but the .SubAddress contains the fragment identifier. When Word encounters this, it once again bravely tries to turn the broken URL into a file link, but to no avail.

This is on 15.0.1512.


Bugs and Issues / Custom table magically appears in locked diagram
« on: February 10, 2020, 08:34:02 pm »
Hi all,

Bit of a weird one. I created a custom table in a diagram, just for playing around with. I created a script in it.

At some point I went to the project browser, located the table element and hit Ctrl-U. EA reported that the table was shown in two diagrams, which I didn't expect. When I went to look, the table I'd created had indeed been placed in that other diagram.

Not by me. And I'm the only one in this project. And the diagram was locked by another user. And the diagram's Last Modified was four days ago. And the table was stretched out to hell horizontally.

So EA dropped my newly-created custom table into someone else's locked diagram.

My question is: is this behavior exclusive to custom tables? Has anyone seen it with other types of element?
I can simply not use custom tables, but if we can't trust EA to maintain the basic integrity of a security-enabled project, then that's going to be a bit of a thing.

I'm on 15.0.1512.


Automation Interface, Add-Ins and Tools / Custom Table Script context
« on: February 10, 2020, 08:16:15 pm »

Anyone had a go at these custom table scripts?

I can create one and get it to execute, but what is its context?
That is, does it have a special variable representing the table? If not, how am I supposed to get at it?
Is there a magic way to interact with the two tags containing the table's format and content?
Does it have access to the full Object Model API? Are there any limitations compared to other scripts?
Can it reference another script? One packaged into an MDG Technology?


Bugs and Issues / Unicode characters when selecting Note column
« on: February 05, 2020, 10:56:59 pm »
Hi all,

I've got a problem with a doc template which uses a custom SQL query to output data from t_object.Note.

Non-English characters in this column are not output as characters in the document, but as character codes (eg &#201; instead of É).
Selecting from the t_object.Name column does not cause this problem. In the schema, t_object.Name is an nvarchar(255) and t_object.Note is an nvarchar(max).

I get the exact same result if I run the query through EA's SQL scratch pad: it works with Name, but not with Note.

There is no formatting in the Note, just regular text.

I'm in a SQL Server repository, using client 15.0.1512 and in the XML Specifications section of my client preferences, the Code Page is set to windows-1252.

What's going on?

More to the point, how do I get it to work?


Pages: [1] 2 3 ... 23