3D Box, after grouping with another 3D Box does not stay in place.

[iglooo101]

http://screencast.com/t/G2sINmD94bq

3D Box, after grouping with another 3D Box does not stay in place.

[iglooo101 (iglooo101)]

I did not add in the JING what happens if there is anything else....each component behaves INDEPENDENTLY even after GROUPING

[Pablo Trabajos (pajarico)]

Can't see the screencast.

[Alvin Penner (apenner)]

similar to Bug 400982 ?

[Alvin Penner (apenner)]

- Windows XP, Inkscape 0.47pre4
- I am unable to view the screencast. However, if I try to reproduce this problem, I get a crash. Attached is a file with two 3D Boxes that have been grouped. When I try to move them I get a crash with the message :

terminate called after throwing an instance of 'Geom::ContinuityError'
  what(): lib2geom exception: Non-contiguous path (src/2geom/path.cpp:355)

[uwesch (uwesch)]

If i try to move grouped 3d boxes under Windows 7 with devel 22522 i get the same error.

[Alvin Penner (apenner)]

as far as I can tell, the crash was introduced some time between
Inkscape 0.47pre1 (Jul 3 2009)
and
Inkscape 0.47pre2 r22153 (Aug 24 2009)

[su_v (suv-lp)]

crash not reproduced with Inkscape 0.46+devel r22528 on OS X 10.5.8
neither moving, rotating or ungrouping the group of 2 3dboxes crashes Inkscape here.

(IIRC there is already another 3dbox bug about crashing when transforming grouped boxes, win32-only as well. This report originally wasn't about a crash AFAIU - but I didn't look at the screencast either. Anyone bother to search for the other report? ;-)

[su_v (suv-lp)]

I was thinking of these reports:
Bug #409043 “3D + rectangle quit after rotation”
Bug #441048 “3D box and circle crash”

both crash with the error message (confirmed on win32 only):
terminate called after throwing an instance of 'Geom::ContinuityError'
  what(): lib2geom exception: Non-contiguous path (src/2geom/path.cpp:355)

… I'm somewhat at a loss how to untangle this net of 3dbox crash and displacement reports! (@Alvin - sorry for my previous comment - I misread your comment how the crash happened ;-)

the new list on win32 (?):
(3dbox + rectangle) + rotate = crash
(3dbox + circle) + translate = crash
(3dbox + 3dbox)grouped + translate = crash

[jazzynico (jazzynico)]

Grouped 3Dboxes dont' crash anymore with 0.47 (tested on Windows XP).
But they still behave strangely...

[Maximilian Albert (cilix)]

Most of the "strange behaviour" mentioned in comment #9 should be fixed as of bzr rev. 8911. Since the crash doesn't occur anymore either, I'm closing this bug. Feel free to open new bug reports with any problems you still experience.

[Michael (null-pointer-us)]

I added nightly ppa and installed 0.47 r9450 and retested grouped 3D boxes for the symptoms described here.

Object -> Group is FIXED
Edit -> Copy is FIXED (my dup #586625)
Edit -> Paste is BROKEN
Object -> Transform and Alignment (on pasted group) is BROKEN

There's still a bug here somewhere.

BTW, thanks for the hard work (towards 0.48); it's looking great!

[Michael (null-pointer-us)]

The "random behavior" seems like operations that are supposed to be applied to the grouped 3D boxes are instead applied to their gradients. Using the alignment tool to center the group horizontally and vertically only centers the gradients (causing color changes) on the 3D boxes, leaving the 3D boxes' positions unchanged.

[su_v (suv-lp)]

> Edit -> Paste is BROKEN

Can you provides steps to reproduce?

> Using the alignment tool to center the group horizontally and vertically
> only centers the gradients (causing color changes) on the 3D boxes,
> leaving the 3D boxes' positions unchanged.

This should be reported separately as regression: transforms using
- 'Object > Transform…'
- 'Object > Align and Distribute…'
- 'Object > Rows and Columns…'
fail on 3dboxes in current 0.47+devel builds (works as expected in Inkscape 0.47).

[Michael (null-pointer-us)]

> This should be reported separately as regression:
> (snipped)

It seems like another part of the same bug. Same objects, same symptoms -- just different steps to trigger it.

But I will file a separate report as suggested.

[Michael (null-pointer-us)]

Ah...nevermind the separate bug report (yet).

I need to clarify something:

The transforms only fail *after* those objects are copied and pasted.

So Edit -> Paste must be fixed before filing bug(s) against the transforms.

> Can you provides steps to reproduce?

Inkscape (nightly) 0.47 r9450

1. Open the attached svg file.
2. Select all objects.
3. Object -> Group
4. (notice bug #300888 is fixed)
5. Object -> Align and Distribute...
6. Click button to (horiz. or vert.) center pasted group on page.
7. (notice how it works as expected)
8. Edit -> Copy
9. (notice bug #586625 is fixed)
10. Edit -> Paste
11. (BUG! objects relative positions and gradients change)
12. Object -> Align and Distribute...
13. Click button to (horiz. or vert.) center pasted group on page.
14. (BUG! transforms to the bugged objects now apply to their gradients)

[su_v (suv-lp)]

On 30/5/10 22:23, Michael wrote:
> Ah...nevermind the separate bug report (yet).

It looks like I made a mistake by marking your initial report as duplicate of #300888: the 'move when copying a group of 3dboxes' issue has been described in the comments of bug #400982 <https://bugs.launchpad.net/inkscape/+bug/400982/comments/10> which probably is more related to yours than #300888. The issues with 3dboxes (groups, c&p, transforms, dragging etc.) seem so intertwined that I was and still am at a loss how to best handle them in the bug tracker.

But as Albert has commented in several of the 3dbox bug reports he recently closed as 'Fixed', «Any further issues should be reported in a new bug report.»

> I need to clarify something:
>
> The transforms only fail *after* those objects are copied and pasted.
>
> So Edit -> Paste must be fixed before filing bug(s) against the
> transforms.

Not quite: a simple 'Paste' _does_ transform the pasted object to the new location under the cursor (if you don't want this, you'd use 'Paste in Place).

The basic new issue IMHO is 'transforming' 3D boxes other than dragging the object with the mouse - interestingly bug #400982 (recently marked as 'Fixed' too) the opposite was mentioned - «the bug shows up only on the mouse release event» <https://bugs.launchpad.net/inkscape/+bug/400982/comments/5>. Could fixing one issue (transforms by mouse) have caused the other to fail (transforms without mouse)?

>> Can you provides steps to reproduce?
>
> Inkscape (nightly) 0.47 r9450
>
> 1. Open the attached svg file.
> 2. Select all objects.
> 3. Object -> Group
> 4. (notice bug #300888 is fixed)

Confirmed.

> 5. Object -> Align and Distribute...
> 6. Click button to (horiz. or vert.) center pasted group on page.
> 7. (notice how it works as expected)

Doesn't work for me (r9456): the group (not yet copy&pasted when following your steps) is already centered on the page, and using other 'Align' commands (top, bottom of page) fails. The result is different depending on whether the outline path is part of the group or not :

a) Group of 3dboxes (pipes and junction): multiple 'align to top' operations (or bottom) don't move the 3dboxes at all, but the gradients appear continuously transformed in the same direction and not aligned to the top or bottom page border.

b) Group with 3dboxes and outline path: using the 'align to top' and 'align to bottom' operations there are 3 states instead of 2 for the outline path (top, middle, bottom) when switching between aligning to top or bottom. The 3dboxes again don't move at all. The gradients don't move endlessly into the same direction but also alternate between 3 different locations.

> 8. Edit -> Copy
> 9. (notice bug #586625 is fixed)

Confirmed.

> 10. Edit -> Paste
> 11. (BUG! objects relative positions and gradients change)

'transform' incorrect. ('Paste' moves pasted objects to a new location under the cursor)

> 12. Object -> Align and Distribute...
> 13. Click button to (horiz. or vert.) center pasted group on page.
> 14. (BUG! transforms to the bugged objects now apply to their gradients)

'transform' fails. (Same as in step 7)

Incorrect/failing transforms of 3D boxes (...

[Michael (null-pointer-us)]

So, do you want me to file a new bug report -- i.e. pick a specific buggy transform to a 3D box, in simple/reproduceable way -- file it, and then see whether any other transforms-to-3d-boxes are still broken once that bug is fixed?

I'm just a user, but IMO these related bugs should be fixed for 0.48 final.

[iglooo101 (iglooo101)]

Guys please notice this bug was filed 2008-11-22 so please do not let this VICIOUS BUG slip into 0.48

[su_v (suv-lp)]

@Michael - sorry again for confusing comments/duplicates: let's move the discussion back to your originally filed report (I will revert the duplicate status) and add your new findings using a recent 0.347+devel build there.