Bug in stitch sub-paths in 0.46 when deleting node

[Steven Devijver]

Attached file demonstrates a bug in Inkscape 0.46. The bug is related to stitch sub-paths and the behavior is Inkscape crashes.

The instructions on how to reproduce the bug are in the file. Basically I combined a four-node square path with a two-node line path and stitched them together. When the left-hand node in the line path is deleted Inkscape gives a message that it has crashed and when I press ok it closes. Deleting the right-hand node works fine.

[mahfiaz (mahfiaz)]

Adding a backtrace from 0.46.
Cannot reproduce this on devel, but on devel the nice red lines show up false, like their nodes all had same y values.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb6244700 (LWP 24756)]
0x0847aa39 in ?? ()
(gdb) bt
#0 0x0847aa39 in ?? ()
#1 0x0845d7f4 in ?? ()
#2 0x0845d121 in ?? ()
#3 0x080de39a in ?? ()
#4 0x080de4cd in ?? ()
#5 0x080debe7 in ?? ()
#6 0x080d7bec in ?? ()
#7 0x080d7cce in ?? ()
#8 0x0858795d in ?? ()
#9 0x08587ea8 in ?? ()
#10 0x08591ce1 in ?? ()
#11 0x0817dff2 in ?? ()
#12 0x0817e156 in ?? ()
#13 0x0817fea7 in ?? ()
#14 0x0817641f in ?? ()
#15 0x08157df1 in ?? ()
#16 0x083b6b66 in ?? ()
#17 0xb7380c4b in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#18 0xb7397095 in ?? () from /usr/lib/libgobject-2.0.so.0
#19 0xb739862b in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#20 0xb792b81e in gtk_signal_emit () from /usr/lib/libgtk-x11-2.0.so.0
#21 0x0839901f in ?? ()
#22 0xb77af036 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#23 0xb737f3c9 in ?? () from /usr/lib/libgobject-2.0.so.0
#24 0xb7380b78 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#25 0xb7396d3d in ?? () from /usr/lib/libgobject-2.0.so.0
#26 0xb739862b in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#27 0xb7398c26 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#28 0xb78c433e in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#29 0xb78d7eff in gtk_window_propagate_key_event ()
   from /usr/lib/libgtk-x11-2.0.so.0
#30 0xb78db1cc in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#31 0xb7f7521c in Gtk::Widget_Class::key_press_event_callback ()
   from /usr/lib/libgtkmm-2.4.so.1
#32 0xb77af036 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#33 0xb737f3c9 in ?? () from /usr/lib/libgobject-2.0.so.0
#34 0xb7380c4b in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#35 0xb7396d3d in ?? () from /usr/lib/libgobject-2.0.so.0
#36 0xb739862b in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#37 0xb7398c26 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#38 0xb78c433e in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#39 0xb77a7c11 in gtk_propagate_event () from /usr/lib/libgtk-x11-2.0.so.0
#40 0xb77a8ef7 in gtk_main_do_event () from /usr/lib/libgtk-x11-2.0.so.0
#41 0xb763f50a in ?? () from /usr/lib/libgdk-x11-2.0.so.0
#42 0xb72e36f8 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#43 0xb72e6da3 in ?? () from /usr/lib/libglib-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#44 0xb72e72c2 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#45 0xb77a93a9 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#46 0xb7eec4c7 in Gtk::Main::run_impl () from /usr/lib/libgtkmm-2.4.so.1
#47 0xb7eec2c2 in Gtk::Main::run () from /usr/lib/libgtkmm-2.4.so.1
#48 0x080864c9 in ?? ()
#49 0x081c5652 in ?? ()
#50 0x080868fd in ?? ()
#51 0xb6793685 in __libc_start_main () from /lib/tls/i686/cmov/libc.so.6
#52 0x08085931 in ?? ()
(gdb) quit

[Alvin Penner (apenner)]

same here, Win32 Inkscape devel version does not crash but the red lines are not useful.
Win32 0.46 crashes but the red lines are editable.

[dopelover (dopelover)]

This problem is still present in devel version (r21453).

[su_v (suv-lp)]

Crash not reproduced on OS X 10.5.8 with
- Inkscape 0.47 r22583
- Inkscape 0.48.0 r9654
- Inkscape 0.48+devel r10014

> but on devel the nice red lines show up false,
> like their nodes all had same y values.

not reproduced in either of above listed versions.

> but the red lines are not useful.

Could you elaborate? The sub-path on the left is a closed sub-path with three cusp nodes with retracted handles, whereby 2 of them on the left side are nearly coinciding.

In 0.47, deleting one and then the remaining node on the left results in a line (where the closing 'z' segment is redundant) and then in a closed one-point sub-path.

In 0.48, deleting one and then the remaining node on the left keeps appearance (with default node tool settings), i.e. the adjacent line segments of the first deleted node are turned into a curve, and in a closed one-point sub-path with extracted handles.

Proposing to close as 'Fix Released' with milestone '0.47' - the oddness lies in the nature of the sub-path, and IMHO is not a bug.

[su_v (suv-lp)]

Tested again:
Crash reproduced with Inkscape 0.46, not reproduced with later releases and current trunk (r10298) on Mac OS X 10.5.8 (i386).

Closing as 'Fix Released' - please reopen if the crash still occurs with the latest stable release (currently 0.48.1).