View source for Sapphire/FlexibleDiagramNodeShapes

This page tracks ongoing issues and discussions pertaining to [https://bugs.eclipse.org/bugs/show_bug.cgi?id=383924 Bug 383924]. 

'''Issues raised from the UI review with our usability expert:''' 

1. <strike>Combo cell editor should be used when direct editing an enum property</strike> [https://bugs.eclipse.org/bugs/show_bug.cgi?id=404500 Bug 404500] resolved 

2. <strike>Remove "direct edit" context menu</strike> [https://bugs.eclipse.org/bugs/show_bug.cgi?id=404503 Bug 404503] resolved 

3. <strike>[Sql Schema Editor Sample] SqlSchema editor sample should include toggle primary key context menu</strike> [https://bugs.eclipse.org/bugs/show_bug.cgi?id=404499 Bug 404499] resolved 

4. <strike>[Sql Schema Editor Sample] Add column action in table node's floating palette doesn't work</strike> [https://bugs.eclipse.org/bugs/show_bug.cgi?id=404501 Bug 404501] resolved 

'''Miscellaneous issues:''' 

1. <strike>Direct editing a newly created node doesn't hide the node label.</strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d4ccb7d0064a970c96dbdf3dc23676e2155bdf79 

----

2. <strike>Select a node that can be resized (ex. sqlschema.xml) and make the node larger.</strike> 

   a. Edit the table label so it is larger, it should have expanded the node to make it fit but does not
 b. Continue to add columns, node does not expand to fit all the newly added columns
 c. Edit the column type which is a combo and it extends outside the node, not utilize the available space.

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d96f45b960d95b7ccf712b589612ce5c39bc8f43 

----

3.<strike>Reorder column (ex. sqlschema.xml) and column inserted in different location than the visual aid (one column below). Works if moving to a column above. Also the visual aid feedback is a few pixel too big. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=bcf33857fb440acb7895a17f3d5f74b6e1957c1f 

----

4. <strike>Drag/drop does not work when moving to the last item - same bug as 3, didn't completely fix. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=7459594a669c7dd7790d61c891cd30fe042d5662 

----

5.<strike>Drag/drop feedback should match orange guide lines</strike>. 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=5ed5db1ff475879c2f8a9e8db4f1e5c9cd1b0a49 

----

6. <strike>UML test case - line figure extends to the right edge, ignored specified margin. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=bb35b4d334ada840adf960f3c06fd4fecfefaca7 

----

7.<strike>Shape Factory always places its children in the 0th index position</strike>. 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=85a274836593b352e977e066e5e38f76ec3d1530 

----

8. <strike>"Show in source" context button not placed correctly in UML text case.</strike> 

<strike></strike>Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=68fe36f36c89455dc59431ae2caec035bdab4037 

----

9. <strike>Shape factory with no separator NPE. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=e18f4d6e5e8e8897247707f037a19133efa27c35 

----

10. <strike>Shape factory -&gt; Case -&gt; Text, note editing not allowed, manual editing in source code results in NPE </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=ea1d0a618962d0f152288624b513f7dea56db9f0 

----

11.<strike>Shape factory -&gt; Case -&gt; Rectangle -&gt; Rectangle -&gt; Validation Marker not displayed.</strike> 

  Workaround: Shape factory -&gt; Case -&gt; Rectangle -&gt; Validation Marker

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=c542a7f75e653d2eb9edc97b7d982c4c82f2e071 

----

12. <strike>sqlschema.xml -&gt; edit the &lt;type&gt;, and type in an invalid text "aaa". Notice the text is still &lt;type&gt;. Type in another invalid text "bbb". Now view the error message, "aaa" is not a valid column type, and notice the error message did not update. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=dc56d000a915a5295b8959dc8047f6aff9a7c342 

----

12.b <strike>atchitecture.xml -&gt; The error message is not shown in the assist dialog. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=3bf509dbe1ee6b77d91878e9b410ab26cbad2dc9 

----

13. <strike>uml.xml -&gt; extra spacing at the end (when the top text margin is 5) </strike> 

Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=879fbe5c5d4588472387ba089306b0423d34f120 

----

14. <strike>uml.mxl -&gt; visible-when on a Rectangle that contains ShapeFactory. First add -&gt; child not added Subsequent add -&gt; ERROR&nbsp;: Figure is not a child java.lang.IllegalArgumentException: Figure is not a child </strike> Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=63e73547f6bc78a9ec494028a7b39ca8ca45a682 

----

15. <strike>sqlschema.xml -&gt; add a table with 2 columns. Give the first column a longer name, edit the second column name. Notice it does not expand to the maximum allowed width.</strike> 

This is not an issue. The sql schema editor's column layout has the column name field left aligned but doesn't expand. The column type field is right aligned and does expand to take up the extra space. So even if the name column appears to have extra space, it is the column type field that has the extra space.<br> 

----

16. <strike>Need a test case for nested shape factories. </strike> 

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=27b96267109590f0285799f149b609fcd1435077 

----

17. <strike>sqlschema.xml -&gt; add a new column to a table in source editor, the new column is always displayed as the last column in the diagram editor even if the new column is inserted in the middle of the table in the source editor.</strike> 

Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=89fd138e02d8d1c1c9c6d46c24b9e505cef69231 

Shenxue: verified 

----

18. <strike>Shape factory -&gt; Case -&gt; Text, the text is editable, selectable and movable, but the selection border is not displayed. </strike> 

Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d6cae2475ac1a9608e6738e70a8102736c30d64a 

----

19. <strike>Added 3 nested Rectangle - each with a different color border and only one of the border is shown.</strike> Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d4949e1c7ddf768ff2243fb20a74fa362aa6f075 

----

20. <strike>In sdef editor, min size, max size, size, expand fields don't show up in the "layout constraint" details page for shapes contained by shape factory case. </strike> 

<strike></strike>Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=27b96267109590f0285799f149b609fcd1435077 

----

21. <strike>Alignment for shapes is only applied when "expand" bit is turned on, making text with default alignment which is center-aligned appear to be left-aligned when there is extra space for the text shape. It's also inconsistent with the cell editor locator for the text figure when direct-editing. </strike> 

Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=5f3fd8c25e08d3d948f01e2180a45af18f6b2b2f 

----

<span style="color:silver">22. <strike> Click away from the validation marker pop up results NPE </strike> </span>

<span style="color:silver">Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=8a84ee0b2a508491160f7a4fb4c877a091a82b6c<br/>
Konstantin: Verified</span>

----

<span style="color:silver">23. <strike>The behavior of the floating toolbar feels odd in the presence of internal selection. Let's make the floating toolbar work off the selection rather than mouse hover. </strike></span>

<span style="color:silver">Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=53975da193eff245eb3e59ebd830d6fe4675d1b8<br/>
Konstantin: Verified</span>

----

<span style="color:silver">24. <strike>Impossible to multi-select and delete internal shapes using the context menu.</strike> </span>

<span style="color:silver">Resovled http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=0a4382698cc60c3aab461110caee898024a114ef<br/>
Konstantin: Verified</span>

----

<span style="color:silver">25. Shift+Click behaves the same as Ctrl+Click when clicking on internal shapes instead of the expected behavior of selecting a range of shapes. (Ling)</span>

<span style="color:silver">Resolved:http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=aefa1f0625be8ba0fb8e42a0d1c4c20b476fd323</span>

<span style="color:silver">25a. The problem still exists in office space editor when selecting occupants.</span>

<span style="color:silver">Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=771ff564e41167ff5e355e7bff5e01dab90a86e6</span>

<span style="color:silver">25b. Still not working as expected in all cases. Trying to shift click a range starting at a field and ending at a method in UML editor sample yielded ctrl click behavior.</span>

<span style="color:silver">Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=1f6b5282c7378d694d6e792dae607950fad4b631</span>

<span style="color:silver">
25c. Still seeing issues. Try it with UML sample and an entity with three fields and two methods with parameters. Trying to shift-select all properties, now doesn't work correctly. I get first and third. Trying to select everything from first field to last method gives me every other item. Also, when shift-click covers a method with parameters, I see method and parameters selected. We should not be selecting a child shape if parent is selected.</span>

<span style="color:silver">Ling: Yes, my last checkin was incorrect. I would like to revert back to Shenxue's last change, in which shift-click stretch select works iff the editpart have the same parent. Otherwise, it reverts back to the jump select similar to ctrl+click. Perhaps there is another algorithm that I've overlooked? <br/>
Konstantin: I think the way to do this is to compute the selection from the geometric perspective. Take the bounds of item 1 and item 2, computing a bounding rectangle, see what selectable parts are intersected and finally remove selectable parts whose parent is selected. You do have to be careful to exclude the selectable parts that fully enclose the bounding rectangle, or you will always end up selecting the node. <br/>
Ling: That was the solution I came up with but had to revert. This breaks down in the uml.xml sample with method1(p1, p2) and method2(param1), shift-click p1 and the param1, p2 is expected to be selected as well, but it is not in the selection bound. <br/>
Konstantin: Why would it not be in selection bound? Since param1 is wider than p1, I would expect the right border of the bounds to be somewhere in the middle of p2, if not convering it completely.<br/>
Shenxue: try this:<br/>
method1( p1 : ReallyLongType, p2 : String ) : String<br/>
method2( param1 : String ) : String<br/>
When p1 and param1 are selected, the combined selection bound doesn't intersect with p2<br/>
Konstantin: I am ok with that. The bounding rectangle idiom is pretty common for handling selection ranges in graphical contexts.</strike><br></span>

<span style="color:silver">Reverted back to the geometric perspective, but added code to selectively remove editparts. It yields surprising results sometimes, but hopefully it is acceptable. Please take a look. 
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=91b51c0545d09a3070f5b7d526c7a97c157d3fc5<br/>
Konstantin: The behavior looks reasonable to me. Verified.</span>

----

<span style="color:silver">26. <strike>Shape factory selection border inset should support negative numbers to expand selection border beyond shape bounds.  <br>
Ling: The negative number is not obvious. There is also the possibility that negative number exceeds the node. I believe the right way to fix uml.xml so that the shape factory children selection is bounded by the parent is to set the parent layout margin to 0 (instead of shape factory->Case->Rectangle layout)<br/>
Konstantin: There is precedent for having positive and negative values when specifying offsets, so I am not concerned about obviousness. We may want to flip the term/property to offset instead of inset, since offset is more common in usage.<br/>
Ling: Unfortunately when I get the paint callback so I'm not able to paint outside the shape's graphic clip rectangle. I don't think it is worth the time and complexity to get this working. Can we mark this as will not fix? <br/>
Konstantin: I can live with lack of offset support on inner shapes, but inconsistency with the node is bothering me. At node level, the current behavior is a 2 pixel offset and I want that to be controllable. Let's do this... (a) rename inset property as offset and flip the sign, (b) support positive or negative offset at node level, (c) support only negative offset at inner shape level (validate for this in sdef).<br>
Ling: The node level inset also has the same issue as the shape factory case, in that paint callback for selection is bounded by the node's graphic clip rectangle. The current selection does not have any offset. But there appears to be a 2 pixel offset for the border. The inset for non-resizable node is trivial and just needs a property editor in sdef. For resizable node, the fix looks more involved. Before I delve into a complex fix, I just wanted to confirm there is a real need. I propose this:
1. offset for shape negative or zero
2. offset for node (negative or zero) enabled iff it is a non-resizable node.<br/>
Konstantin: Ok. Let's do offset for shapes (negative or zero) and leave node with no offset support for now (hardcodded behavior). I am sure that we will need to come back to this, but this should be good enough for now.
</strike>
Resolved by changing inset (positive or zero) to offset (negative or zero) only. 
http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=7168447ea27759582ada810fc8cff2314422c11d<br/>
Konstantin: Verified</span>

----

<span style="color:silver">27. <strike>sqlschema.xml drag &amp; drop column results in ClassCastException. </strike> </span>

<span style="color:silver">Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=e5a8ffda80ffc0032a0d0b817911faab8a2cb9c8<br> Konstantin: Verified</span>

----

<span style="color:silver">28. <strike>When in direct edit mode, pressing a tab should move editing to the next field.</strike></span>

<span style="color:silver">Resolved by http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=8cc53e15b678a8a0c706529d3d931774cb7cd2ef and</span>
<span style="color:silver"> http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=70129132e6912ac34c1efeb7645b704006d357ec</span>

<span style="color:silver">Note: Pressing tab moves the editing to the next text field in the node, or the first text field in the next adjacent node. When determining the next adjacent node, all the nodes are sorted using the x, y coordinates of the top left corner,  from top to bottom and then left to right<br/></span>

<span style="color:silver">Konstantin: Verified</span>

----

<span style="color:silver">29. <strike>NPE at DiagramDeleteActionHandler.java:100 sometimes when deleting shapes in UML editor sample. </strike></span>

<span style="color:silver">Ling: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=0753d12a83f8f4aaa2397a4bb5b94b5be91d9abc <br/>
Konstantin: Verified </span>

----

<span style="color:silver">30. <strike>Default text color should be black. Use "black" rather than "#000000".</strike> </span>

<span style="color:silver">Resoved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=378a147d6e0f679be5907cf367fc916b8cc6e8dd<br/>
Konstantin: Verified</span>

----

31. Need to be able to specify Properties View Contribution for a shape factory case. Currently, can only be done at node level.

Shenxue: Since there is no part for shape factory case, the Properties View Contribution is built into shape part. I've augmented sdef editor to allow developers to specify properties page for shapes. http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=3025c8a6a40808a68d4aba9a17e08b577bad32c0<br/> 
Konstantin: It works, but I don't like how it works. It's too hard to tell where it is actually legal to contribute to the properties page. At least at sdef level, properties view contribution should be on node and shape factory case, not on shape.

----

<span style="color:silver">32. <strike> Need to be able to control selection presentation for the node, similar to what's possible for a shape factory case. Additionally, the background of the selection should be controllable in both cases. The default should probably be no background. (Ling)
</strike>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=c75fbd984296aa46b499436d5c39a69e96fb5922 <br/>
<strike>
Konstantin: I don't see selection presentation controls for a node in sdef editor, nor do I see selection background control for shape factory case.<br>
Ling: I just rebuilt and verified that they are there. For nodes, the selection presentation options are under Nodes -> entity -> Shape. For shape factory, the selection background section is below selection border under Shape factory -> Case.<br/>
Konstantin: Ok. I see it now. Must have been looking at a stale build. But why is inset not available at node level?<br/>
Ling: Didn't implement an inset for node because the nodes might be resizable. Seems weird that the resize feedback is different than the actual resize. <br/>
Konstantin: I expect inset/offset to work for nodes. Right now, it looks like a hardcodded 1px offset. </strike><br/>
Ling: Resolved - see 26. </span>

----

33. <strike>Delete all bendpoints action shows up when selected multiple shapes. (Ling) </strike>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=611e32524cb8608d7f6b71a2406bcfb018528ae8 <br>

<strike>33a. "Delete All Bend Points" menu does not disable/enable when adding/removing bend points. <br>
33b. "Delete All Bend Points" menu is always disabled when select multiple connections.</strike> Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=41a00345af5c2a1d4175ebbce151388de72f409f <br>
33c. <strike>"Delete All Bend Points" menu does not update when selection is the same while modifying bend points.</strike> Resolved:
http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=81f6f4e8a776a648269f2213dfff561637fd909e

----

34. In the layout, cell sizes (min, max and preferred) should be after margins are applied, rather than before. 

Ling&nbsp;: Not a bug - this is consistent with the SWT GridLayout<br> Konstantin&nbsp;: Consistency with other layouts is not a goal.<br> Ling: Yes, but I suspect most of Sapphire users are experienced with SWT GirdLayout and would find the new way to cell size calculation surprising.<br> Konstantin&nbsp;: Let them be surprised. The goal is to take good ideas from existing layouts and leave the rest behind, so if you are going to argue for the current behavior, convince me that it is the better semantics. Consistency arguments do not sway me. <br> Ling: Let say the user wants to specify an image size of 32x32 and margin of 2. Let say after testing the margin needs to be increased. Currently, the user changes the margin to 3. But the new way the user needs to modify both the margin to 3 and the preferred image size since the size is inclusive. I also think it is easier to think about adding the margin plus cell size, vs, total size minus the margin. <br> Konstantin: The size settings reference the size of the cell, not the size of what's contained. Note that in your scenario, there is no reason for the user to be specifying any size settings. Just the margins will do. <br> Ling: Agree that setting size should be minimized. Let say you want to specify image size in conjunction with visible-when. You are specifying the actual size of the image, not the size plus the margin. <br> Konstantin: Visible when makes the whole cell disappear, so size settings and margins are not relevant. <br>
Ling: There may not be a reason for the user to specify any size setting, but if they do specify a size setting and a margin for an image, then their image will be clipped by the margin. I fail to articulate what seems so obvious to me. <br>
Konstantin: It only seems obvious because you are used to incorrect usage of the word margins in SWT GridLayout. Think about the meaning of the word margin. My document is 8.5"x11" with 1" margins, not 7.5"x10" with 1" margins. CSS is a better reference point. They have margins and padding. Margins are on the inside. Padding is on the outside.<br/>
Ling: Yes, you are right. But there is some validity in "when in Rome, do as the Roman's do." I'm probably not the only person brainwashed by SWT margin and confuse margin with padding. <br/>
Konstantin: The layout we have is so drastically different from anything available in SWT that I am not concerned by this. If anyone is confused, they will quickly get unconfused through a bit of trial and error.

----

35. <strike>Spacer part is needed. This part should be treated as a transparent elastic, capable of size from 0 to infinity. All sizing-related layout constraints should apply. Besides the layout constraint, the only user-configurable property should be visible-when.</strike> 

Shenxue: Resovled http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d36035cd612678630a8b7bc07d02ac3086dfcb92<br> 

----

<span style="color:silver">36. <strike>Validation marker part should handle lack of a validation problem by making the part invisible. Override SapphirePart.initVisibleWhenFunction().</strike> </span>

<span style="color:silver">Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d36035cd612678630a8b7bc07d02ac3086dfcb92<br/>
Konstantin: Verified </span>

----

<span style="color:silver">37. <strike>Image part should handle lack of an image by making the part invisible. Override SapphirePart.initVisibleWhenFunction to augment this implicit constraint to whatever the user may have specified.</strike> </span>

<span style="color:silver">Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=d36035cd612678630a8b7bc07d02ac3086dfcb92<br> Konstantin: I don't see ImagePart overriding initVisibleWhenFunction(). Also, there is something weird going on in ImagePart. I am guessing that imagePathFunction and getImagePath() aren't actually needed. <br/>
Shenxue: Based on a later discussion, part visibility and null image/text should be handled separately. I've verified the current code handles them correctly. </span>

----

<span style="color:silver">38. <strike>ShapeFactoryPart does not implement computeValidation. It needs to rollup the validation result for the parts that it creates/manages.&nbsp;</strike> </span>

<span style="color:silver">Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=77c456619b043c44276aa27716f5be7b9d071b1d<br> Konstantin: <strike>This only partly works. When new child element is added or removed, the parent validation marker is not updated. Try adding and removing method parameters in the UML sample.</strike> Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=17890341599643ef91932e31e76a9bcf19148275<br> Konstantin: Verified </span>

----

<span style="color:silver">39. <strike>Separator part from a shape factory is selectable. It should not be. You can see this in the UML editor. If you are very precise, you can select the coma between method parameters.</strike> </span>

<span style="color:silver"><strike>Ling: This was resolved in an earlier checkin. <br/>
Konstantin: I am still able to repro the issue.</strike><br/>
Resoved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=96895122b78354d41228302e43ebce75301853b8<br/>
<strike>Ling: Introduced ClassCastException </strike><br/>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=a2557e7936be868ff742dbfa95bc8d922c630564<br/>
Konstantin: Verified</span>

----

<span style="color:silver">40. <strike>Show in source is missing from the context menu of the shape factory case.</strike> </span>

<span style="color:silver">Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=14c0c2ce327e13e97d43948d43b15b95916b9f82<br> 
Konstantin: Verified</span>

----

41. <strike>In a stacked layout, the first shape controls the dimensions of the stack. This should be ok in many cases, but is inconsistent with how other two orientations work. Expect the dimensions to be established by taking all of the participating shapes into account.</strike>

Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=c85d3406a52c9ccfcfa755ef2fcf55d499913a68 Above commit was reverted. <br>

In our stack layout implementation, the first shape controls how the subsequent controls are placed. Thus it makes sense that the first shape controls the dimensions.<br/>
Konstantin: The first shape controls nothing. In fact, the first shape may change over time given various dynamic behaviors, such as visible-when. The placement is always relative to the cell, not the first shape, and cell size should be calculated taking all shapes into account.<br/>
Ling: This change will not be backward compatible. In the architecture sample, the validator and depends on image's layout constraint is relative to first shape control.<br/>
Konstantin: Not sure I understand. Backwards compatible to what? If the first shape is largest, the behavior would be unchanged from today.<br>
Ling: architecture sample will no longer work, as well as task flow in OEPE. This only an issue because the cell is wider than image width.<br/>
Konstantin: I don't understand why you are saying that these usecases will not work. If you are saying that they will need tweaking of layout constraints, that doesn't concern me.<br/>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=71b345c47184550dd6bff07afc5e3f589e4b18d9<br/>
Konstantin: Something is still not right with the layout. Looking at the architecture sample, the inner rectangle should not be expanding, but it is. Since the base image is the largest of its children, I expect the inner rectangle to stay pinned to the size of the base image plus margins. The adjustments of the overlay margins should not have been necessary. Let's talk more about this tomorrow.<br/>
Shenxue: The architecture sample is designed to have more room for the node label thus the cell size for the image is larger than the image size. The first image(base image) is center aligned horizontally and vertically so there is some padding at the left and right. To place the other images (decorators), it's much easier if the margin is calculated based off the first image compared to margins based off the cell. In order to calculate the margins of the decorators based off the cell, we have to know how large the cell is and how the base image is placed (centered vs left-aligned etc). <br/>
Shenxue: more fixes in http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=3487416b7e818b1f0694461d42706f086ea495c5. Please give it a try and let me know if you see issues there.<br/>
Konstantin: It's much better, but I still found issues. Pull down my changes before looking at these... (a) margins on inner rectangle aren't equal (4 pixels on left/top and 5 pixels for right/bottom), <strike>(b) node with setting overrides shape's capabilities, I set node width to 200 and expand horizontally on both rectangle and text to false, the node width was still 200, expect it to be within the capabilities of the shape.<br/></strike>
Shenxue: regarding issue b, there is no size constraints set on either the rectangle or the text. Therefore the allowable max size for them is Integer.MAX_VALUE. The outer rectangle's allowable max size is also the max integer.  If a max size constraint is set on a child, e.g., see http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=ccbecb0b3a0d85522f3db3150ef2bd4fa4f6832e, you'll notice the node width of 200 is overridden.<br/>
Shenxue: per more discussions on this size constraint issue, I've checked in more fixes http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=2e4f2dd5194c8837fc4b74c5a9b768a43d49d2e6 and http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=225cd61d254fac3994b6e632c2961fca365ebcd1. Issue b should be considered resolved.<br/>
Konstantin: Still not working correctly. I set expand horizontally to true for text, but left it false on the inner rectangle. The node size expanded to 200 specified at the node level. All children have to allow the expansion to happen.<br/>
Shenxue: I'm really confused now. For a vertical sequence layout, the max width of the parent is the max width of all children. If one of them is allowed to expand, its max width is max integer. That allows its parent's max allowable width to be max integer. If some children are allowed to expand, some are not, how should we calculate the max allowable width for the parent? I think our size specification and constraints are really confusing. If I explicitly specify node size, I'd expect that size to be respected as long as it's bigger than the min size of children and at least one of its children is allowed to expand. This gives user fixed node size layout and I'd think there are many use cases for it. <br/>
Konstantin: The max width of the parent is not max width of all children. It's the maximum width that all children allow. In other words, its min of max width of all children combined with max of min width of all children, since you aren't allowed to clip anything.

----

42. <strike>Gradient background - select only 1 gradient background get NPE. Also sdef UI should check for less than or equals to 2 gradient segments.</strike>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=fd67ae549885a11b928006d5ef4130d0d2f4491b

----

43. <strike>Node -> Shape -> Text, NPE when direct edit. (Ling)</strike>
Resolved: http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=2f128ecd2c23a5fee7ba0621d3f616d55014e061

----

44. Node selection outline seems to have one pixel padding on top/left and no padding on bottom/right. (Ling)<br/>
Ling: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=9c681085e6ec7fdb53fff1552affb0d98139c91d<br/>
Konstantin: Verified that the node selection outline spacing from the border is consistent (0 px on all sides), but there is a new issue. If you look at the UML editor sample, there is a pixel missing from the node border in upper-right and lower-right corners.

----

<span style="color:silver">45. <strike>The image shape should allow developers to implement a double-click action. This can be used to implement simple editing behaviors, such as cycling through enum or boolean values when the image is the representation of such properties. In the UML editor, I would like to implement double-click for entity kind image (cycle between class and interface) as well as for field/method accessibility icon. Let's call this action "Sapphire.DoubleTap", since "tap" is more generic than "click" in the days of touch screens. Note that I don't expect Sapphire to provide any action handlers for this action, just the action, so that developers can implement handlers. </strike></span>

<span style="color:silver">Shenxue: Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=cc7f1e67f46b5c12a28d210bd378927a48b228fc<br/>
Shenxue: I also added an action handler for the entity kind image to toggle the entity type in the uml sample.<br/>
Konstantin: Verified</span>

----

46. <strike>sql schema editor sample, create a new table, resizing the table node, save the diagram, reopen the diagram, the table node is not displayed.</strike>

Resolved http://git.eclipse.org/c/sapphire/org.eclipse.sapphire.git/commit/?id=b811105073e9f47754f000e73caeccfb252e171d

----

47. sqlschema sample, resize a node, then start adding columns making sure the new size is bigger than the resized node. Save and close. Reopen and the node size is the resized size and all the added columns are not visible (until resized)