Regression problems on closed Spiro (rev >= 14226)

[Jabiertxof]

Not sure when, maybe in last 2Geom Sync,we have problems on closed paths with spiro:
1 problem: SpiroLive crash when close path backtrace: https://inkscape.org/en/gallery/item/6231/
2 problem: Apply spiro to a closed path, no crash but not work and log:
** (inkscape:4167): WARNING **: Exception during LPE Spiro spline execution.
 lib2geom exception: Non-contiguous path (../../trunk/src/2geom/path.cpp:990)

[su_v (suv-lp)]

Please add information about Inkscape build (at least the revision number) to the bug report.

[su_v (suv-lp)]

Attaching externally hosted backtrace to the bug report (external links tend to expire or get removed over time).

[su_v (suv-lp)]

On OS X 10.7.5 (using archived builds as well as latest trunk r14241):

1) crash when closing spirolive path:
- not reproduced with Inkscape 0.91+devel rev <= 14224,
- reproduced with Inkscape 0.91+devel rev >= 14228;
very likely a regression introduced with the 2geom update in rev 14226:
http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/14226

2) dysfunctional spiro path effect when applied to closed path:
- not reproduced with Inkscape 0.91+devel rev <= 14224,
- reproduced with Inkscape 0.91+devel rev >= 14228;
very likely a regression introduced with the 2geom update in rev 14226:
http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/14226

[Krzysztof Kosinski (tweenk)]

This problem can be fixed by calling setStitching(true) on the affected path.

[Jabiertxof (jabiertxof)]

Hi Krzysztof thanks with your help. I try to fix but no luky :(
Spiro code is so complex to me, I try to make some Stitching changing originals curves to paths to allow add stitching but no works.

[Krzysztof Kosinski (tweenk)]

Does r14250 fix the problem?

[Jabiertxof (jabiertxof)]

No this not fix the bug. the same backtrace.
../../trunk/src/2geom/path.cpp:987 no check for stiching value, could be this?

[Alvin Penner (apenner)]

attached is a proposed patch for this bug.

in the original code, if the spiro path was closed, the code would add a 'close path' command immediately after the first 'move to' command, which caused a discontinuity. In the new code, the 'close path' command is postponed until after the last segment is drawn.

Any testing would be welcome.

[Jabiertxof (jabiertxof)]

Thanks Alvin for the patch, I try it tonight!!!

On vie, 2015-08-07 at 13:37 +0000, Alvin Penner wrote:
> attached is a proposed patch for this bug.
>
> in the original code, if the spiro path was closed, the code would add a
> 'close path' command immediately after the first 'move to' command,
> which caused a discontinuity. In the new code, the 'close path' command
> is postponed until after the last segment is drawn.
>
> Any testing would be welcome.
>
> ** Patch added: "spiro_seg.diff"
> https://bugs.launchpad.net/inkscape/+bug/1473641/+attachment/4441060/+files/spiro_seg.diff
>

[Jabiertxof (jabiertxof)]

Hi Alvin tested and go very well, no console wrong output and work as expected. The code show great! ¿Is too much hard fix this bug? I try to fix previously but no luck :(
Thanks for the patch!

[Alvin Penner (apenner)]

glad to hear it is working. The problem is not really coming from 2geom, it is isolated to just the spiro code, but it was accidentally exposed when the definition of continuity in the routine Path::do_append was slightly changed.

in any event, I will wait for a day or so, and if no other comments show up, or no problems are encountered, then I'll commit the change. Thanks for testing!

[Jabiertxof (jabiertxof)]

You are welcome!

[Alvin Penner (apenner)]

fix committed to rev 14285