jonkiparsky / tree-buildingsurvey Goto Github PK

What steps will reproduce the problem?
1. Enter text entry mode in an empty node
2. Hit "delete" or "left arrow". You may have to enter text, then use the
left arrow to leave some text in the box while you delete off the left side. 

Result: The button bar disappears. There is a lot of console activity. 
Typing any text or hitting right arrow restores button bar and stops
console activity. 

From the console, this line seems significant: 


Exception in thread "AWT-EventQueue-45" java.lang.IllegalArgumentException:
Zero length string passed to TextLayout constructor

1. Create a new node
2. Start labeling it
3. Try to move it

What is the expected output? What do you see instead?

Should be able to move it around. But I can only move it up and down, not
right and left.

4. Go to a different tab in the browser
5. Return to the applet window

Now I can't move the new node at all - nor can I type in the label, despite
the blinking cursor there.

(I can make it disappear by creating a new empty node, as in previously
submitted bug.)

1. Connect to test applet
2. move stuff from the palette, move stuff around

What is the expected output? What do you see instead?

Some organisms are duplicated rather than simply moved. 

I will attach a screen shot on the wiki since I don't seem to be able to do
that here.

1. Log in as Ethan
2. Answer the last of the survey questions
3. Log out
4. Log in again, see congratulatory message
5. Proceed to work on tree

What is the expected output? What do you see instead?

I expected to see what I'd left behind - not much. Instead I saw about 10
empty nodes in random places.

I'm noticing an aggravating interaction between the current (0.3) applet
and the mouse/keyboard handling in the google code site. Aggravating
because it's difficult to nail down the conditions, which seem to be quite
specific. 

So far, I've reliably reproduced this by double-clicking to create an empty
node, and giving a "yes" response to the dialog box that asks if I want to
enter text. I enter text, and the node is created, but...
...if I have a google code tab open in the same window, all of a sudden
mouse/text handling is kerput. I can't select a text entry field, if I
click on text it wants to be dragged somewhere instead of giving me an
insert point, weird stuff like that. I can select text, and the keyboard is
still active (if I'm in an Issue, typing "s" toggles the star for that
issue, for example) but I can't tab from field to field and typing does not
enter text or replace selected text. Selected text can be moved.  
Closing the window and restarting ends the problem, and it doesn't affect
other sites that I've found.
I suspect that this is a simple problem, and it might indicate a need for a
stop() method. 

What steps will reproduce the problem?
1. Connect two Nodes
2. Drag either across the other. 


Behavior on screen is normal, but console output shows a string of output
beginning with:

Exception in thread "AWT-EventQueue-1" java.lang.NullPointerException
    at tbs.TBSUtils.getConnectionBounds(Unknown Source)
    at tbs.view.TBSView.renderConnections(Unknown Source)
    at tbs.view.TBSView.paintComponent(Unknown Source)
(...)




This is in version 0.4, rev 405, running on MacOS X - please try to
reproduce on your home machines and verify this console output. 

What steps will reproduce the problem?
1. Start up v0.4, connect some organisms

When an arrow is under construction its tail jumps around on the source
node in a confusing manner -sometimes it starts from the middle, sometimes
from a corner. I think it should always start from the middle (usually top
or bottom, possibly left or right).

In a finished link, the arrowhead appears at the end of the arrow. That's
confusing when two arrows have the same organism as target. Perhaps
directions should be indicated this way --->--- rather than as ------->.
There might still be a problem when arrows cross, but it'd be less likely
to cause confusion.

You might want to check the second of these with Prof White.

In the open-response answer fields, a word which runs across the maximum
line length is split, rather than wrapping to the next line. 

When a saved tree is restored, text in open-answer boxes is not wrapped at
all. Long lines continue to the end of the box and disappear. 

What steps will reproduce the problem?
1. Add objects: Crocodile, Bird, Rat.
2. Press Undo 3 times

What is the expected output? What do you see instead?
First two Undo presses remove Rat and Bird. 
Undo still shows active button.
Third press of Undo, Crocodile stays in place and Undo button becomes
inactive. 


Further symptoms: 
Clear does not work after this happens. Pressing Clear calls confirmation
box, but pressing "Yes" does not dismiss box, and pressing "No" returns to
same tree.

Also, Object can be Selected, but Delete button and Delete key do not work. 

 * Object can be deleted manually, by dragging off screen. 
 * If another object is added at this point, both can be deleted using
either Delete Button or Delete key. 






When only one object is showing on the screen, the Undo button still shows
itself as enabled. 

Please use labels and text to provide additional information.

Console output at "undo" of first action (add Crocodile to tree):

Exception in thread "AWT-EventQueue-8" java.util.EmptyStackException
    at java.util.Stack.peek(Stack.java:85)
    at tbs.model.StudentModel.removeFromTree(StudentModel.java:503)
    at tbs.model.history.Add.undo(Add.java:22)
    at
tbs.controller.StudentController.handleMouseButtonPressed(StudentController.java
:524)
    at tbs.controller.StudentController.mousePressed(StudentController.java:232)
    at java.awt.Component.processMouseEvent(Component.java:6345)
    at javax.swing.JComponent.processMouseEvent(JComponent.java:3255)
    at java.awt.Component.processEvent(Component.java:6113)
    at java.awt.Container.processEvent(Container.java:2085)
    at java.awt.Component.dispatchEventImpl(Component.java:4714)
    at java.awt.Container.dispatchEventImpl(Container.java:2143)
    at java.awt.Component.dispatchEvent(Component.java:4544)
    at java.awt.LightweightDispatcher.retargetMouseEvent(Container.java:4618)
    at java.awt.LightweightDispatcher.processMouseEvent(Container.java:4279)
    at java.awt.LightweightDispatcher.dispatchEvent(Container.java:4212)
    at java.awt.Container.dispatchEventImpl(Container.java:2129)
    at java.awt.Component.dispatchEvent(Component.java:4544)
    at java.awt.EventQueue.dispatchEvent(EventQueue.java:635)
    at
java.awt.EventDispatchThread.pumpOneEventForFilters(EventDispatchThread.java:296
)
    at
java.awt.EventDispatchThread.pumpEventsForFilter(EventDispatchThread.java:211)
    at
java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:201
)
    at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:196)
    at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:188)
    at java.awt.EventDispatchThread.run(EventDispatchThread.java:122)

1. Create a node
2. Select label mode
3. Click on created node
4. Enter some text
5. Try to delete

What is the expected output? What do you see instead?

I expected the delete key to delete. Neither delete nor backspace does
that. Backspace adds unprintable characters. Delete deletes the whole node.

I'm not sure this will reproduce the error, but here's what I did.

1. Create three new nodes, A, B, C
2. Link A to B.
3. Link A to B again.
4. Try to delete the link from A to B.
5. Try to delete the link from A to B again.

What is the expected output? What do you see instead?

I expected the first delete to delete both links. When that didn't happen I
thought perhaps one was left. But the second delete did nothing.
So ...

6. Link B to C.
7. Try to delete the link from B to C.

THE LINK(S) FROM A TO B WENT AWAY!
Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Place two ONodes on the screen.
2. Connect first to second. 
3. Select first (to get connection line)
4. Click and drag second.

What is the expected output? What do you see instead?
Expected output should be revert to null/null - attempt to connect to an
object that is already connected should fail. 

This reveals flaws in the current mouse semantics of connection; these
should be nailed down and fixed in place.
I'm scheduling this for version 0.6 fix.

What steps will reproduce the problem?

Cannot provide specific steps to reproduce. 

I built a tree, including a bat, which was linked to a number of other
objects. 
As I was manipulating the tree, I periodically performed a "save" and
"restore" (using the perl-generated save button, not the one in the
applet's button bar)

Sometimes, when I did the save and restore, the bat was in the lower right
hand corner. All of its links were preserved, and it was always the bat. 

What steps will reproduce the problem?
1. create empty node
2. add label
3. go elsewhere

What is the expected output? What do you see instead?

Label should be "complete" but I still see a blinking text entry cursor there. 

That's bad enough. The reality is worse. If I then create another empty
node and try to label _it_ the first node I created disappears! I can
reproduce this behavior.



Please use labels and text to provide additional information.

1. Do some stuff
2. Select delete on the toolbar
3. Click on one item
4. Click on another

What is the expected output? What do you see instead?

Both items were deleted. I expected just the first one to vanish.

I think the normal UI model for deletion is that you select something and
then tell the system to delete it - with the delete key, or control-x, or
from a menu. Your app reverses this - you select delete mode and then
decide what you want to delete.

I hope you can convince Prof White to change the semantics. If not, I think
you should turn off delete mode after the first deletion.

Finally, I needed to stare at the screen to see that delete mode was still
selected. The color change on the toolbar button was not very visible.

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Add an Organism Node to the tree
2. Drag the Empty Node on the left (immortalEmptyNode) over an Organism Node

What is the expected output?
No Empty Node should be created.
Empty Node on left should still be there

What do you see instead?
Instead, left Empty Node disappears

Instructions say 

# Shift-click on the workspace to add a node

but that doesn't happen.

1. Click on an organism


Should show visually that it's been selected. Sometimes it has been
sometimes not. The only way to tell seems to be to move the cursor away
with the mouse and see if there's an arrow following.

What steps will reproduce the problem?
1. Create some empty nodes
2. Double click on them
3. Select "Yes" when asked if you want to delete them

What is the expected output? What do you see instead?
Empty nodes should always be deleted after the user selects yes, but
sometimes they are not

Dialog boxes for creating empty node do not show any text, only
confirmation buttons, in MacOS X.
I'll look into it. 

What steps will reproduce the problem?
1. Logged into http://cluster.bio.whe.umb.edu/cgi-bin/TreeSurvey.pl with 
universla password
2. Applet did not display any organisms in IE7 browser


What is the expected output? What do you see instead?

I am assuming we are supposed to be seeing a list of organisms to set up a 
tree for.

What steps will reproduce the problem?
1. Move an organism to the right
2. Move another
3. Click on the first to start an arrow
4. Try to finish the arrow at the second organism

What is the expected output? What do you see instead?

Nothing I do seems to allow me to terminate the arrow. It even follows me
if I drag the first organism back to the left, in hopes that deleting it
will delete the arrow I've started.

If I play around, clicking and double clicking, eventually something
happens. Once because I double clicked it made a the connection to the
second organism and then started a link from that organism ...


Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Dragging node places it at edge of screen, as close to drop point as
possible.
2. Redo step one with another organism


What is the expected output? What do you see instead?
Organisms should not overlap. They sometimes do. 

Please use labels and text to provide additional information.

1. Play around a bit
2. Click on undo

What is the expected output? What do you see instead?

I expected the last action to be undone. Instead I saw a picture of the
horseshoe crab appear on the screen. Moreover, the label on the left, which
I think should have been left behind, just disappeared. It's a good thing I
knew what a horseshoe crab looked like.

1. Select label button
2.
3.

What is the expected output? What do you see instead?

Instructions say "Please being typing"

I think you mean "Please begin typing"

The TBS applet as currently executed does not load on all browsers and
systems. We believe this is a consequence of the novel method of storing
student tree information to Professor White's server. 

At this time, the applet loads on Windows consistently. It loads in certain
Linux configurations and fails in others, and it refuses to load on Mac OSX
under all standard browsers and most non-standard ones, with iCab being the
sole exception found to date. (Safari, Firefox, Chrome, Omniweb, and Camino
all failed). 

What steps will reproduce the problem?
1. Open up any of the versions in a browser
2. Javascript warning exists in bottom-left corner "Object Expected" Line 
18, Char 1

What is the expected output? What do you see instead?

No Warnings

1. Create an empty node
2. Drag it to the left of the window

What is the expected output? What do you see instead?

It should be deleted. Instead it seems to remain at the left. 

Play around with creating and deleting empty nodes this way and you will
see several kinds of strange behavior.

(I suspect the fix is to get the semantics right, not to tinker with cases
in the code. Deleting an empty node (with or without a label) should just
make it go away forever. Deleting an organism node from the tree by
dragging it to the left restores it to its place in the original list.
Somehow these outcomes are confused or conflated.)

Created multiple organisms pointing to one Empty Node. Unlinked ENode, then
selected undo. Connection arrows reappeared, but did not track motion of
either target node or source node. 

Odd button functioning when this happened:

Delete, select connection: deleted previously selected object (connection
shows selected, darker green)

Delete, select an object: deletes object and connection

Unlink, select an object (with one connection, frozen) unlinked object.

There is probably further odd button behavior to explore here. 

What steps will reproduce the problem?
1. Drag an empty node on top of an organism node

What is the expected output? What do you see instead?
Collision avoidance should reposition empty node.
It doesn't. 

Due to the way connections are structured, as directional objects, we
currently allow (have not yet disallowed) a bi-directional connection: Ant
to Crocodile and Crocodile to Ant. Is this a mistake we should allow the
user to make, or a possibility we should eliminate?

If we allow it, which I'm inclined to do, we need a means to distinguish
the two connections. In the arrowhead condition they will be visually
distinguished by the arrowhead, but selecting one or the other connection
will be difficult. 
Would it be posssible to require that a return connection adhere to a
different corner of one or both objects?

What steps will reproduce the problem?
1. Connect two organisms. 
2. Drag one over the other

What is the expected output? What do you see instead?
When one organism is over the other, the connection line points to the
upper left corner. 

What steps will reproduce the problem?
1. Open the [http://cluster.bio.whe.umb.edu/cgi-bin/Test/TBSTestSurvey.pl 
current version]
2. Move an organism into the tree
3. Hover the mouse over that organism

What is the expected output? What do you see instead?
The tooltip displaying the organism's name should display much closer to 
the node, but it is dislaying about three/four times further away from the 
node

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Move a label
2. Press "undo" 



What is the expected output? What do you see instead?

Expected output: Label is replaced in origin of move.
Actual output: new copy appears at origin of move, but old copy remains in
place. 

There may be other odd behavior in "undo" but none reproducible yet. 

What steps will reproduce the problem?
1. Create an Empty Node
2. Drag an Organism Node over it

What is the expected output? What do you see instead?

Organism Node should snap back to last good position,
but this does not occur

When labeling, delete key doesn't delete but does leave an unprintable
character in the label. 

Backspace key does delete.

Note: can't edit a label except by backspacing from the end and retyping. I
expected a cursor (in a textfield).

In the course of working up the previously reported bug (Text entry/losing
focus), I managed to lose the applet altogether. I believe the last button
I hit was Undo and the applet disappeared, leaving a blank space where the
applet had been drawn. For now, this is visible to anyone who logs in as me. 

I don't know the precise sequence of steps, but there was a lot of shifting
windows, a lot of save and restore, and a lot of undoing. 
I will try to replicate the steps if our database guru can remove the
broken tree (might want to save it if possible, as useful data)

The console output on restarting is attached, in case anyone finds it useful. 

What steps will reproduce the problem?
1. Connect two nodes
2. Delete one
3.

What is the expected output? What do you see instead?
Connection should be deleted. It isn't. 

What steps will reproduce the problem?
1. Run the applet. Move the mouse. Do anything

What is the expected output? What do you see instead?
Expected: things move and exist without a lot of flicker. 
Instead, things flicker. A lot.

When I return to a previously saved tree, the model shows a no-go on the
"unlink" button, even if there are links. Clicking "unlink" reports "can't
unlink, and we stay in Select mode. Selecting an object with links does not
change the status of the Unlink button, but clicking on the ("disabled")
button unlinks the object. 

All source files need the GPL header boilerplate as a comment.

1. drag some organisms to tree
2. click on link
3. see message
4. link as directed
5. continue work

What is the expected output? What do you see instead?

I expected the help for link to disappear when I'd made the link. But it
hangs around.

Returning to a saved tree with empty nodes in it, I expect the "Label"
button to show itself as enabled, since there are empty nodes to be
labelled. Instead, it is x-ed out. 

Clicking on the disabled Label button does not allow selection of an empty
node for labelling. 
However, clicking on an existing empty node, followed by Label, allows text
entry in that node. 

Normal behavior of the Label function is restored when an Empty Node is added. 

What steps will reproduce the problem?
1. Create an Empty Node.
2. Use the "Label" button to enter text entry mode, and enter some text.
3. Without hitting return, select another browser window. 
4. Return to applet. 

First problem: The empty node remains yellow, and the flashing cursor
remains visible, but the node is no longer the text entry point. 

5. Click in the left-hand area of the empty node and drag the mouse on a
diagonal. 

Second problem: The purple Empty Node will move on the horizontal axis and
the yellow text entry box will move on the vertical axis. 

[deleted issue]

I've posted for consideration a pretty major change to TBSModel - please
diff the two versions, and see which one is easier to work with. 
I think you'll see the point of the changes, let me know if there's a
better way to get these effects. 

Applet does not load from MacOS. Not the package issue this time. 

What steps will reproduce the problem?
1. Click and drag on an object.

The object moves as expected but does not retain focus. (is not highlighted
and is not subject of next operation)



Please use labels and text to provide additional information.

General comments:

The messages are too wordy. It's nice to be polite but making the user
parse "please" and "simply" over and over again is tedious. Just say what
needs to be done.

The messages use the same screen area as the tree, so they interfere with
nodes at the top of the screen. Think about a better place to put them.
Perhaps they can just pop up on mouseover. That way the experienced user
doesn't need to see them all the time.

1. Click on latest applet link on wiki home page


What is the expected output? What do you see instead?

I expected to see the latest version. I got version 4, and wasted time
playing with it, entered a new bug, then found out by visiting the schedule
that 0.4 was not the latest!