Open GoogleCodeExporter opened 9 years ago
Can someone give a detailed descrption what happens?
The platform isn't able to reload lasad. Lasad is shown in an iframe like all
other tools, and we can not change that fact. If it happens if the lasad tab
was invisible when the object was shared and become visible afterwards this
issue can be connected to the hide method for tabs
(http://dev.sencha.com/deploy/gxt-3.0.0/javadoc/gxt/com/sencha/gxt/core/client/S
tyle.HideMode.html). We use at the moment the hide mode HideMode.VISIBILITY and
this is necessary to avoid reloading of the other microworlds after the tab is
changed.
Original comment by irgang...@googlemail.com
on 3 Jul 2012 at 12:54
TD any thoughts what to do here?
Original comment by mavrikis@gmail.com
on 3 Jul 2012 at 2:45
Not sure I understand the problem but one can Override the GWT onLoad method to
have any code run when a widget is finally displayed.
Original comment by ToonT...@gmail.com
on 4 Jul 2012 at 2:08
Could this be related to Issue 139? Regardless, from what Ken says above we
could call some code to re-center the map or something because the current
situation is quite problematic and cognitively overloading for the students.
Not sure who should be the owner here.
Original comment by mavrikis@gmail.com
on 5 Jul 2012 at 6:36
Sorry I haven't gotten to investigate this yet, I will today. I could imagine
that it is related to issue 139, since some XMPP message makes the server put a
loaded tool into a strange state (this doesn't happen when loading LASAD
outside the platform and sharing an object).
However, I don't think the issue is that the onLoad method is getting called again, as, from my understanding, that would show the loading bar and look as if you're restarting LASAD, not just shift your view.
I will look into this more today and report anything I find.
Original comment by tdragon1...@gmail.com
on 5 Jul 2012 at 6:46
I wasn't suggesting it was getting re-loaded.
onLoad()
This method is called immediately after a widget becomes attached to the browser's document.
http://google-web-toolkit.googlecode.com/svn/javadoc/latest/com/google/gwt/user/
client/ui/Widget.html#onLoad()
This will be called when the widget or its container is for example made
visible or added to a visible container. onAttach is better name but the
documentation for that suggests one use onLoad.
Original comment by ToonT...@gmail.com
on 5 Jul 2012 at 7:34
Sorry, I was confusing this with 'onModuleLoad' which is called only when
loading the first time (I believe).
This onLoad method, does it have some default behavior? Is your theory that it
is called each time the user switches tabs to display LASAD again within
Metafora?
I don't think it solves the bug, but it could be a solution that I call to
re-center the map each time with this onLoad method, if it is automactially
called at the appropriate time. The question remains, what happens from
outside LASAD (in Metafora) that makes the map scroll-bars reset? I'm
investigating now.
Original comment by tdragon1...@gmail.com
on 5 Jul 2012 at 7:45
I believe it is called when tabs are selected. Try adding this to an
appropriate extension of the Widget class:
@Override
public void onLoad() {
super.onLoad();
System.out.println("onLoad called.");
}
Original comment by ToonT...@gmail.com
on 5 Jul 2012 at 8:20
I have a feeling (but not well tested) that this may happen only in Firefox
(similar to the juggler issue that requires chrome). I couldn't test now nor am
I sure that this changes anything really as should work in Firefox.
It doesn't happen outside the platform (so this is something to do then with
the Metafora tab widget?)
Original comment by mavrikis@gmail.com
on 6 Jul 2012 at 11:45
Here's my possibly incorrect understanding. When a tab is selected all the
widgets of that tab are attached to the DOM. When another tab is selected those
widgets become detached. If, while detached, some JavaScript code runs (e.g.
generated by GWT) that asks for coordinates or dimensions of detached widgets
they get zero for a reply.
If this is right, then the two most obvious workaround are to store the
coordinates when attached (onLoad) and use those rather than getAbsoluteLeft()
and the like. Alternatively fix or delay the layout when onLoad is called.
Just my guess...
Original comment by ToonT...@gmail.com
on 7 Jul 2012 at 1:08
This issue happens more often than just when obejct is shared, and seems to
happen to other tools as well (Planning tool map disaapears each time an object
is shared, or accessed. Combining here with another issue about the same
problem:
- Open a discussion resource card first time
- Takes you to lasad in the centre
- Go back to planning tool
- Click on the card again (without closing lasad)
- It takes you to top left of the screen.
Original comment by tdragon1...@gmail.com
on 10 Jul 2012 at 7:37
Changed the title then from "Sharing an object from expresser shifts lasad map
to top corner." to a more general one (feel free to change again)
Original comment by mavrikis@gmail.com
on 10 Jul 2012 at 7:49
Issue 105 has been merged into this issue.
Original comment by mavrikis@gmail.com
on 10 Jul 2012 at 7:50
So we can't do much for the current period so I removed the critical label and
the LKL mileston.
That's a more general issue so passing to KUEI.
Original comment by mavrikis@gmail.com
on 10 Jul 2012 at 2:14
From a technical view this Issue is the same as Issue 139 and Issue 125. The
problem is caused by the hide method. At the moment we use hide mode visibility
which uses the visibility css style attribute. The hide mode display is no
option because this one cause a reload of the Unity Microworlds. I think the
offsets hide mode also caused some problems.
We will test the offset hide mode next week and I hope we will find a solution
for this problems.
Original comment by irgang...@googlemail.com
on 12 Jul 2012 at 3:52
Issue 125 has been merged into this issue.
Original comment by mavrikis@gmail.com
on 12 Jul 2012 at 4:28
Issue 139 has been merged into this issue.
Original comment by mavrikis@gmail.com
on 12 Jul 2012 at 4:28
I added a milestone and raised this to critical (it means it is critical for
that milestone --- in this case the conference presentations). This makes it
easier (I think) to manage issues (happy to talk if you prefer another way of
managing them)
Original comment by mavrikis@gmail.com
on 16 Jul 2012 at 1:03
We have tested all GXT hide modes and no one solved the issue, but this issue
only affects planningtool and lasad with the firefox browser.
Original comment by irgang...@googlemail.com
on 26 Jul 2012 at 5:12
I also experience this problem with chrome, when moving to Juggler, saving
state, and then returning to the planning tool.
Original comment by tdragon1...@gmail.com
on 29 Aug 2012 at 9:39
Since this problem is prevelant (happens often) and problematic (difficult for
users, looks bad for demos), and we dont' seem to have any pending solutions, I
wonnder if it's not time to consider workarounds.
One suggestion (from me) is to add a small "reload frame" button somewhere,
maybe just blow the tabs in the top-left corner. Then, at least you could
quickly click it rather than needing to: right-click->this frame->reload frame.
Original comment by tdragon1...@gmail.com
on 29 Aug 2012 at 9:42
This issue is not related only to hide / show functions, because it happens to
others' computers when one person changes something. For example:
-user 1 and 2 are looking at the planning tool
-user 1 opens a juggler resource card
-user 2 loses their plan view, has to reload frame.
I just produced this twice in a row, using both chrome and firefox (one user
each) on the same computer.
Original comment by tdragon1...@gmail.com
on 29 Aug 2012 at 9:50
Also to note, reloading effects others as well.
-user 1 and 2 are looking at the planning tool, both blank due to eralier loss
-user 1 reloads frame
-user 2 also has their frame refreshed
Again working in chrome and firefox on the same machine
Original comment by tdragon1...@gmail.com
on 29 Aug 2012 at 10:30
As of 2012.09.07, I now have to reload twice to bring the planning map back.
In firefox on mac
Original comment by tdragon1...@gmail.com
on 7 Sep 2012 at 12:02
Original issue reported on code.google.com by
mavrikis@gmail.com
on 28 Jun 2012 at 9:14