Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 88568 - [ViewMgmt] Sticky views don't appear in perspective or appear when not expected
Summary: [ViewMgmt] Sticky views don't appear in perspective or appear when not expected
Status: VERIFIED FIXED
Alias: None
Product: Platform
Classification: Eclipse Project
Component: UI (show other bugs)
Version: 3.1   Edit
Hardware: PC Windows XP
: P3 major (vote)
Target Milestone: 3.1 RC1   Edit
Assignee: Kim Horne CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-03-19 12:38 EST by Konrad Kolosowski CLA
Modified: 2007-02-15 10:59 EST (History)
5 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Konrad Kolosowski CLA 2005-03-19 12:38:57 EST
There is some weird logic with opening sticky views and saving and restoring 
perspective state.  I was surprised many times by Help view appearing out of 
nowhere after restarting the workbench.  Try these scenarios.


Working Scenario:
1. Have more than one perspective open.  Ensure Help view is closed in all 
perspectives. Restart the workbench.
2. Press F1.  Help view appears in the perspective - OK
Switch to another perspective, Help view present - OK.

Not Working Scenario:
1. Have more than one perspective open.  Close Help view in all perspectives. 
Restart the workbench.
2. Browse through all open perspectives
3. Press F1.  Help view appears in the perspective - OK
Switch to another perspective, Help view not present - Problem.

4. While on the perspective without Help view, restart the workbench.  The 
perspective still does not have the Help View.
5. Switch back to the perspective that has help view, restart the workbench 
like in step 4.  Switch to the perspective that did not have the Help view, it 
now shows the Help View - Problem, it is a surprise as the result is different 
from step 4.

6. Close Help View in one perspective.  Switch to another perspective.  Help 
view is still there.  I believe, for predictable behavior, sticky views should 
appear AND disappear simultaneously across all perspectives.
Comment 1 Nick Edgar CLA 2005-03-21 10:50:15 EST
Kim, would you be able to investigate?
Comment 2 Kim Horne CLA 2005-05-04 12:51:44 EDT
I will try and investigate for M7.
Comment 3 Konrad Kolosowski CLA 2005-05-17 12:16:22 EDT
Raising to major.
It is important to the usability Help view.  Sad to not see it already fixed 
in M7.  Without it Help view is annoying.  It is designed to deliver 
assistance unobtrusively, and this bug makes the experience opposite.  Open 
help view once and you are plagued by it when you don't need it.
Comment 4 Kim Horne CLA 2005-05-24 09:54:06 EDT
I've improved the sticky state logic and made the result persistant across sessions.  The issues you've 
raised should be addressed now, but perhaps others have been introduced.  By my estimation things are 
OK but I'd appreciate someone at your end (the help end) take a look at this and see if the behaviour is 
adequate for 3.1.
Comment 5 Nick Edgar CLA 2005-05-24 11:39:49 EDT
Kim, if a perspective is closed and then reopened, do you need to reset the state?
In the current change, I don't see any handling of the close case.
Comment 6 Nick Edgar CLA 2005-05-24 11:49:01 EDT
Reopening pending answer.
Comment 7 Kim Horne CLA 2005-05-24 13:38:09 EDT
Actually, it was cleaning up after itself in disposePerspective but I did break that during the most recent 
checkin.  Fixed in HEAD.
Comment 8 Dejan Glozic CLA 2005-05-26 18:19:56 EDT
Konrad, are we now happy with the sticky Help view behaviour?
Comment 9 Konrad Kolosowski CLA 2005-05-26 20:37:44 EDT
The definition of sticky view "... will appear by default across all 
perspectives ..." may be interpretted in various way, so we could argue 
details for very long.  I will not.

I have not encountered surprising appearance of help view out of nowwhere in 
recent builds when switching perspectives, so I am satisfied for 3.1.
Thank you.

Comment 10 Dejan Glozic CLA 2005-05-26 21:10:19 EDT
Konrad, the original spec for a sticky view was to have the 'feel' that the 
entire perspective stack has been pushed aside to make room for the view. This 
ways, perspective changes should have no effect on the view. 

I don't know what happens now when the view is closed in one perspective. That 
aspect is less clear to me - I remember asking for the view to go away from 
all perspectives but I don't think it happens like that today.
Comment 11 Konrad Kolosowski CLA 2005-05-26 22:07:55 EDT
Dejan,
I see opening the view shows it in all other perspectives that never had it 
(in workspace life time).  From that point on, if you close the view in some 
perpspective, and open the view, the view will open only in the current 
perspective.  That essentially makes the sticky view work like a regular view 
now.  There is not much stickyness about it.  This behavior does not justify 
having the fast view artifacts at all, but if this behavior ships in 3.1, we 
can use it for help view, there will be some minimal benefit to the new user 
in the first few minutes of using the product.

Closing the view does not remove it from all perspectives, and without this 
implemented, the today's opening behavior is probably preferable.  The view 
gets introduced, and is not annoying.

Comment 12 Kim Horne CLA 2005-05-27 08:40:26 EDT
Closing a view in a non-visible perspective is actually quite a bit of work.  

Seeing as you're happy with the behaviour I will mark as verified.
Comment 13 Jamie Liu CLA 2007-02-14 11:18:20 EST
Based on usability tests for our product, users are getting confused when they see the help view reappear when they change perspectives, after having closed the help view.  They are expecting that when they close it in one perspective, it closes it in all other open perspectives, not just the active perspective.  This behavior is described in #6 of the original reporter's bug report.  Please consider reopening.
Comment 14 Konrad Kolosowski CLA 2007-02-14 11:34:03 EST
Hi Jamie,
There were multiple, pretty bad, bugs with sticky view implementation that were all fixed.
Point 6 is not a code defect.  It is problem with the definition of the sticky view, and I do not think it is reasonable to reopen this defect.

I am not sure it is appropriate to change definition, and affect all sticky views.  It might be better to introduce improved sticky view, called differently, or having an additional attribute in its contribution such that it clarifies behavior across perspective.  Consider opening a separate feature request, that addresses only that point.
Comment 15 Matthew Hatem CLA 2007-02-15 10:59:08 EST
Please see related bug #174213.