Thanks for the update and the work. I'm glad to see the regression is still available in trunk.
Maybe applying the regression would be easier than trying to fix the code?

-C

On Sun, Mar 20, 2016 at 12:56 PM, Jabiertxo Arraiza Cenoz <jabier.arraiza@...2893...> wrote:
Is a bug "know" bug.
https://bugs.launchpad.net/inkscape/+bug/1507049

Thanks for the info!


El dom, 20-03-2016 a las 11:58 +0000, C R escribió:
> I re-installed 0.91 stable release, and opened the same file
> generated by
> inkscape-trunk, to see what would happen.
>
> As expected, Inkscape 0.91 (r13725) interperets the offset correctly.
> I
> jiggled the offset handle a little so Inkscape detected a change,
> then
> re-saved the file.
> Here is the re-saved result for comparison (open in browser to see
> the
> difference):
> http://www.opendesignstudio.org/inkscape/inkscape_dynamic_offset_0.91
> _stable.svg
>
> Also, here is a gif animated comparison of the old offset and new on
> the
> same shape:
> http://www.opendesignstudio.org/inkscape/dynamic_offset_old_new_compa
> rison.gif
>
> -C
>
>
>
>
> On Sun, Mar 20, 2016 at 10:21 AM, C R <cajhne@...400...> wrote:
>
> >
> >
> > You have an idea how much time could this bug happen?
> > >
> > >
> > Unfortunately, it happens on every shape I've tried. It's 100% of
> > the
> > time, as far as I can tell.
> > Before, I got perfect shapes every time.
> > Now, all the vectors have either a slightly corrupt, or very wavey
> > and
> > corrupt result after applying the offset.
> > It's almost as if the algorithm is trying to apply a smooth effect
> > or
> > something.
> >
> > The result should be the same as applying a very thick line to the
> > shape,
> > then converting stroke to path, and deleting the interior of the
> > stroke.
> > This effect must be perfect. It should not have to guess or
> > estimate
> > anything, and should be pristinely uniform.
> >
> > Whatever was done to the calculation of the offset needs to be
> > reverted
> > back to the way it was.
> >
> > Thanks again for your help, Jabier.
> >
> > -C
> >
> >