Re: [Lib2geom-devel] Bezier roots problem

[jf b. <jf....@gm...>]

hi all!

I still have no time to play with 2geom, but I'm happy to read news from it
:-) !

I don't know exactly what the state of (sbasis) boolops in 2geom is, but I
did work on it. At the end, I was relatively happy with the topological
stuff, but I bumped into an unexpected "bug" iirc:
subdivision of sbasis is not exact, even at order 0.
If you split an sbasis at some point 0<t<1, the two pieces don't match
exactly. They do most of the time, but sometimes they only do up to a tiny
little micro error. Note we cannot expect the path not to "move" when doing
the subdivision (and the topological approach was ok with that), but the
two ends should match.

I wanted to fix that, but I don't remember if I did indeed. The easiest way
would be to force exactness by hand. We could force it up to some order of
derivatives. We could even compute all the relevant derivatives and rebuild
the two sbasis from that. This would slow the code down a little
unfortunately...

With this fixed, I think the code was not too bad, and robust as long as
our root finding system is. I remember I wrote some toys but I did not take
the time to clean anything up, so the code may be a little crappy. Does
boolops LPE use something like that?


I still hope to find a little time for 2 geom, but I'm affraid it wont be
before a long time if ever: I would like to revive the sbasis of 2
variables and the computation of $\int f(x,y) dy$, in order to do harmonic
(= nice envelope-) deformations... I wish I had more lives.


Bye, JF.


2012/1/8 Johan Engelen <jbc...@sw...>

> Hope you have a nice vacation!
>
> When you come back, you can play with Inkscape's boolops LPE to find
> bugs in the code. At the moment, it simply hangs until it hits a memory
> limit (goes into infinite loop I think). It used to work, if buggy.
>
> Ciao,
>   Johan
>
>
> On 5-1-2012 7:14, Nathan Hurst wrote:
> > On Wed, Jan 04, 2012 at 09:21:04PM +0100, Johan Engelen wrote:
> >> I have 2 beziers that are mirror images of each other:
> >> Bezier(0, -20.8271, 20.8271)
> >> Bezier(20.8271, -20.8271, 0)
> >>
> >> roots(0) gives 2 solutions for the first, but only one for the second.
> >>
> >> /src/tests/bezier-test also fails for roots...
> > booga.
> >
> > I'm on holidays in hawaii atm, will look next week hopefully.
> >
> > njh
> >
>
>
> ------------------------------------------------------------------------------
> Ridiculously easy VDI. With Citrix VDI-in-a-Box, you don't need a complex
> infrastructure or vast IT resources to deliver seamless, secure access to
> virtual desktops. With this all-in-one solution, easily deploy virtual
> desktops for less than the cost of PCs and save 60% on VDI infrastructure
> costs. Try it free! http://p.sf.net/sfu/Citrix-VDIinabox
> _______________________________________________
> Lib2geom-devel mailing list
> Lib...@li...
> https://lists.sourceforge.net/lists/listinfo/lib2geom-devel
>