search

eclipse-uml2 / uml2

An EMF-based implementation of the UML 2.x metamodel for the Eclipse platform.
Eclipse Public License 2.0
5 stars 4 forks source link

BIDI3.3:HCG_Path variable is corrupted when define complex variable name #24

Open eclipse-uml2-bot opened 2 weeks ago

eclipse-uml2-bot commented 2 weeks ago

| --- | --- | | Bugzilla Link | 182453 | | Status | NEW | | Importance | P3 major | | Reported | Apr 15, 2007 08:58 EDT | | Modified | Oct 05, 2010 16:17 EDT | | Version | 2.1.0 | | Depends on | 160682 | | Reporter | amir |

Description

Build ID: 20070323

Steps To Reproduce:\ Please notice that this is a BiDi specific defect and it can be reproduced only when BiDi data is used. Please use BiDI characters instead of capital Latin letters in the data samples below.

Launch eclispe in mirrored mode (using -nl iw attribute)

  1. Open a UML model: Select File > New > Other... menu item. Select the UML Model wizard from the Example EMF Model Creation Wizards category and press the Next > button
  2. Enter a name for the uml model
  3. Select the Advance>> button
  4. Check the "Link to file in the file system" checkbox.
  5. Click on the Variables... button
  6. Click on New... button
  7. Type in a complex variable name contains Hebrew and English characters. For instance: ABCabc (this is the typing order)
  8. Click on Folder... button
  9. Select some path for the varaible
  10. Click OK.

Result:\ Actual Result: The path variable is displayed wrongly in the path variable dialog: abc - C:\mypathCBA\ Expected results: C:\mypath - abcCBA\ (Where capital letters represent Hebrew characters)

More information:\ See attachemnts for illustration

eclipse-uml2-bot commented 2 weeks ago

By amir on Apr 15, 2007 09:05

Created attachment 63832 Path variable dialog with two kinds of variable names

path_variable.JPG

path_variable.JPG

eclipse-uml2-bot commented 2 weeks ago

By James Bruck on Apr 27, 2007 13:04

Based on some research by Ed this will be deferred until the next release. See comments below...

I've tried to chat with Steve about this, but he's not followed up recently. And Cam was also promising to send a note, but that never happened. Clearly everyone is too busy with other things.

I've carefully read the small amount of documentation available and it's very clear to me that this problem is still wide open and that the resolutions, if any, are poorly understood. For example, a label provider doesn't know what orientation its labels will be used in, and hence doesn't know what directional characters need to be inserted. Also, when we fetch a message, we don't know if the message will be translated, so we don't know whether we need to add characters to force a particular orientation. The problems are many, the solutions (if you can call them that) are poorly understood and not so accurate, while the experts remain in short supply or out of touch. So I would suggest that none of these will be resolved in this release cycle, and that no real user will be bothered by that fact.

If this were really a serious problem, my efforts to get involved and help arrive at a solution would not have run into dead ends so many times. So assuming this is important, it's something that we should start looking at in depth at the start of the next release cycle, not at the end of this one.