XMI Export

General Questions/Discussions
1

Hello,

XMI export doesn’t keep polish national characters and replaces them to sequences of type \uXXXX, e.g.:

rodzaje p[b]ł[/b]atno[b]ś[/b]ci

is replaced with:

rodzaje p[b]\u0142[/b]atno[b]\u015b[/b]ci

It causes some problems with exporting to MDA tools.

Regars,
Jacek

2

Hello again,
There is another issue - when exporting to XMI with UML2 option, multiplicities * (for roles or attributes) are represented as:

When importing model to eclipse they are recognized as 1.
So I have to use 0…* to have correct multiplicities, that are represented as two elements:

There is no semantic difference between * and 0…* and * is shorter. Maybe * should be represented the same way, as 0…*

Regards,
Jacek

3

The third issue - when importing to eclipse, elements haven’t stereotypes and tagged values, assigned to them in VP.

4

We have observed this same problem-- it is because the application of stereotypes in UML2 is handled through profiles, which VP doesn’t yet support, if I understand correctly. I’ve been unable to import XMI files defining profiles (authored in Eclipse) into VP-UML, for the same reason, I believe.

Jick, could you please clarify? Are there plans to add profile support to VP, in accordance with the latest versions of UML2? We deal with stereotypes and tagged values a great deal and it would be nice not to have to come up with another mechanism in order to get full “round-trip” compatibility.

Thanks,
Ellen

5

Hi Jacenty and Ellen,

Thank you for the posts. We will investigate the problems and get back to you.

Best regards,
Jick

6

Hi Jacenty and Ellen,

This has been fixed, but the fix will be available in the coming version. Do you want to try it? If yes, I can arrange an early access release for you.

Best regards,
Jick

7

Hi Jick,
I’ll be busy next week, but then I’d like to try it.

Thank you,
Jacek

8

hi Jacek,

The next version may be released by the end of the next week. :slight_smile:

Take your time.

Best regards,
Jick

9

Hello,
We have the same problem, when exporting the XMI with french characters :
The “é” charcter is replaced in the xmi by “\u00e9”.
Is there any correction of this issue available.

Many Thx
Regards,

10

Jick,

Thanks for the info. I would also be very interested in evaluating the fix ASAP-- please keep me posted!

  • Ellen
11

Hi Riadh and Ellen,

In the coming version, we will let user select an encoding for the XMI, which will solve this problem.

Best regards,
Jick

12

Hi,
Thx for your replay. Is there any patch correcting this issue. i understood reading the forum that it’s an old issue.
May we know when this new version will be released. and be posted if possible.
Many thx
Regards,

13

Hi Riadh,

I regret that there is patch available for this fix. However, this will be available in the coming version, which will be released very soon. We expect to roll out the new version by the end of this week.

Best regards,
Jick

14

Hi,
many Thanks.

reagrds,

15

Jick,

Thanks for the information on the upcoming release. Will this include the fixes to the Eclipse UML2 export as well? (That’s the main issue that would prevent us using VP-UML at this time.)

Thanks,
Ellen

16

Hi Ellen,

Sure. The fix will cover UML2 Export as well.

Best regards,
Jick

17

Jick,

I’ve downloaded the evaluation copy of 6.1, but I’m sorry to say I’m still not seeing the desired functionality when it comes to UML2 exporting. Perhaps I’m doing something wrong?

Please take a look at the attached zip file. The .vpp within was initially created by importing a simple model (in UML2/XMI format) from Eclipse. (The model is based on one from the UML2 Tutorial.) I altered it by creating a new Stereotype (simply called <> ), then I applied this stereotype to the “Address” class. This worked perfectly within VP-UML.

The .uml file is the result after the model was exported from VP-UML (UML 2.1, “UML2 compatibility” selected). Whereas TestStereotype is still defined within the model, it is no longer applied to the Address class. This was the behavior I was hoping would be present in the new version.

Can you shed any light on this? Perhaps I’m laboring under a misapprehension.

Thanks,
Ellen
VP61StereotypeTest.zip

18

hi Ellen,

Sorry, I haven’t mention that it is only the first and the second problem get fixed. For the third problem, which is the one you spoke of, we are still investigating.

Best regards,
Jick

19

Hi Jick,
There are any issues in new release:

  1. When I use default encoding “windows-1250” and generate xmi file, polish characters are changed with something like escape sequences, but if I choose “other” option, and then “windows-1250” from the combo box, it works well.

  2. When importing generated file with eclipse, following error message occurs: Declared encoding “windows-1250” does not match actual one “Cp1250”; this might not be an error.
    Import works well after changing the encoding tag from “windows-1250” to Cp1250.

Regards,
Jacek

20

Hello Jacenty,

We will investigate the problems. For point 2, may I ask you for a sample XMI so that we can reproduce and fix the problem faster? Thank you.

Best regards,
Jick