Pattern along Path broken in 0.91, both on-canvas and on pdf/png export

Bug #1428267 reported by Amadiro Fafarello
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Inkscape
Fix Released
High
Alvin Penner

Bug Description

pattern-along-path (the path effect, not the equivalently named extension) is broken in 0.91. This regression first occurs in 0.9x, in 0.48, pattern-along-path was rendering correctly.

Exported PNG image of the broken behaviour:
https://dl.dropboxusercontent.com/u/21559589/p-o-p-broken.png

Exported PDF of the broken behaviour:
https://dl.dropboxusercontent.com/u/21559589/p-o-p-broken.pdf

SVG file that exposes the broken behaviour:
https://dl.dropboxusercontent.com/u/21559589/p-o-p-broken.svg

There are a variety of different control point configurations in which the erroneous rendering occurs, it appears to me that it happens most of the time when AB (the line between the first node and its handle) and CD (the line between the second node and its handle) are almost but not exactly parallel.

This makes pattern-along-path harder to use, and means many SVG files that use p-o-p that opened correctly in 0.48 are not displayed correctly in 0.91 anymore. On most of my more complex SVG files which use pattern-along-path heavily, about 20-40 paths expose bad rendering and need to be manually tweaked to sidestep the problem.

The incorrect rendering seems to be unaffected by zoom-level, rotation or scaling of the path (using the F1 tool)

Sometimes the bug occurs in a very obvious manner, sometimes it is very subtle (which makes it very devious, as you don't tend to catch it before after the high-quality printing process...)
Here are some more examples of the bug in action:
https://dl.dropboxusercontent.com/u/21559589/p-o-p-bug-1.png (a very obvious one)
https://dl.dropboxusercontent.com/u/21559589/p-o-p-bug-2.png (nearly every one of the small pathes here exposes the problem.)
https://dl.dropboxusercontent.com/u/21559589/p-o-p-bug-3.png (an obvious one)
https://dl.dropboxusercontent.com/u/21559589/p-o-p-bug-4.png (a less obvious one on a small object)

su_v (suv-lp)
tags: added: livepatheffects
ScislaC (scislac)
Changed in inkscape:
status: New → Confirmed
importance: Undecided → High
su_v (suv-lp)
tags: added: regression
Changed in inkscape:
milestone: none → 0.92
Revision history for this message
ScislaC (scislac) wrote :

Reproduced with 0.91.0. It appears that it's partially fixed in the 0.91.x stable branch (r13770) and trunk (r14151). Attaching modified test file which still shows the issue still exists. ~suv reported that this file looks as expected in 0.48.x (r10043).

Revision history for this message
su_v (suv-lp) wrote :

Based on tests with archived builds (on OS X 10.7.5), and the original test case:
- not reproduced with r12920,
- reproduced with r12921;
this regression seems to be related to the 2geom update in r12921:
http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/12921

Other regressions introduced with r12921:
Bug #1272119 “trunk: poor rendering of path effects after 2geom update (rev >= 12921)” (fixed)
Bug #1312333 “trunk: incorrect rendering of circle segment (rev >= 12921)” (fixed)
Bug #1428683 “occasional bad behaviour of envelope deformation LPE (rev >= 12921)” (fixed)

The fix for bug #1428683 in r14005 AFAICT helps with the reporter's original test case (not ok with r14004, ok with r14012), but not with the one provided by ScislaC.

Changed in inkscape:
status: Confirmed → Triaged
Revision history for this message
Amadiro Fafarello (jwringstad) wrote :

Thanks for looking at the bug.

I can confirm that the bug occurs to a somewhat lesser extent in 14151, although in certain situations it does return in full force. Many of the cases that had extreme spikes now only have mild (but noticable) deformations, perhaps that provides some clues:

https://dl.dropboxusercontent.com/u/21559589/p-o-p5.png
https://dl.dropboxusercontent.com/u/21559589/p-o-p6.png

In each case the deformation goes away when nudging the node handle slightly in any direction, as per usual.

I can also confirm that I still get the broken behaviour with the "p-o-p-broken+.svg" that ScislaC posted.

Revision history for this message
Alvin Penner (apenner) wrote :

thanks for the heads-up, I'll take a look at it

Revision history for this message
Alvin Penner (apenner) wrote :

fix committed to rev 14161

su_v (suv-lp)
Changed in inkscape:
assignee: nobody → Alvin Penner (apenner)
status: Triaged → Fix Committed
tags: added: backport-proposed
Revision history for this message
ScislaC (scislac) wrote :

Backported fix to 0.91.x r13771

Changed in inkscape:
milestone: 0.92 → 0.91.1
tags: removed: backport-proposed
Revision history for this message
Amadiro Fafarello (jwringstad) wrote :

I have just recompiled the latest trunk (Inkscape 0.91+devel r14194 custom) and can confirm that the regression does not occur anymore with any of the documents it used to occur with.

jazzynico (jazzynico)
Changed in inkscape:
milestone: 0.91.1 → 0.92
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.