#9 Open path fixes

[Ruwan Janapriya]

Hi Timo,

As we have discussed in a previous ticket, there is a (or few) bug(s) in open paths. I've forked your code and incorporated some of the related commits from C# code base. Is there a way to contribute your project, for a new release with latest changes (including those open path fixes). Specially, are there any tests other than demo code to make sure new changes do not dbreak anything?

This is where the changes I've added located
https://github.com/japzi/jsclipper

The change log / or commits
https://github.com/japzi/jsclipper/commits/master

Demos
https://japzi.github.io/jsclipper/index.html?p=open_closed.html

Thank you!

Ruwan.

[Ruwan Janapriya]

Hmm, couldn't find a way to edit the original ticket.

More changes or commits are available here, in 'easel' branch
https://github.com/japzi/jsclipper/commits/easel

Thank you

Ruwan

[Ruwan Janapriya]

Hi Timo,

Do you think we could add these fixes to the original code? Atleast the open path fixes?
Also, how do you test your code before a release, just to avoid regression?

We are using your awesome library in www.easel.com

Thank you

Ruwan.

[Timo Kähkönen]

I have no timetable for adding new fixes. I assume that all those bugs are fixed in C# Clipper newest version and I have an intention to make use of the same code as in C# Clipper.

Angus has also not been capable of making updates, and I have a bit waited the release of floating point version of Clipper. It is huge speed improvement. You can test my beta-version:
http://jsclipper.sourceforge.net/6.1.3.4b_fpoint/main_demo3.html

http://jsclipper.sourceforge.net/6.1.3.4b_fpoint/clipper_unminified_6.1.3.4b_fpoint.js

You can of course make fixes and use the fixed version and release it somewhere if you need.

[Timo Kähkönen]

The time of making 1503 operations:

FPoint: 1836 ms
IntPoint: 5688

FPoint:
http://jsclipper.sourceforge.net/6.1.3.4b_fpoint/main_demo3.html

IntPoint:
http://jsclipper.sourceforge.net/6.1.3.2/main_demo.html

If using big integer range, the speed gain is much more bigger.

[Ruwan Janapriya]

These are awesome numbers. Are there any known issues in FPoint library? I understand it is beta, just curious how stable it is?

[Timo Kähkönen]

I have not found any issues when testing with about 100 000 operations/polygon combinations. There are some differences between IntPoint and FPoint versions in common line joining (attached image). I'm not sure about which version produces "correct" result. This is one thing why I'm in hold regarding updates. I have a feeling that both versions doesn't make correct result when polygons with common lines are rotated (like in the JS Clipper Main Demo, attached image).

[Timo Kähkönen]

I made an update, where those open path bugs are fixed or should be at least:
https://sourceforge.net/projects/jsclipper/

[Ruwan Janapriya]

Thank you very much!!! I will be testing this today!!

[Ruwan Janapriya]

Tested and it works great! Thank you!!

[Timo Kähkönen]

Fine! I'm waiting Angus Johnson to make updates. He wrote at the Sourceforge page of original Clipper:

17 March 2016:
Since it's been some time since the last update, some may be wondering if I no longer plan further updates. I do have plans for more updates but I also have a chronic health condition (not life threatening) that's set me back, and I don't really have a good idea when I'll restart development.

The thing that I'm waiting is the Floating point version, because it is so much faster at least in Javascript.