[patch] cosmetic fix for KPlotWidget

Hans Meine hans_meine at gmx.net
Sun Dec 2 19:45:58 GMT 2007


On Freitag 30 November 2007, Jos Poortvliet wrote:
> > The real bug is that integer coordinates do not denote the centers of
> > pixels. (Think of it this way: square pixels do not exist; they are
> > simply the nearest-neighbor interpolation of the sampling points, which
> > are at integer coordinates.)  This makes live hard for people (like me,
> > doing high-accuracy image analysis) who care about sub-pixel positions
> > and such "neglectable" offsets.
> >
> > To become OT again: Jason's patch is the right workaround for this bug in
> > Qt AFAICS. ;-p
>
> You call this a bug - would that mean as soon as this is fixed we're
> gonna see fuzzy lines again?

I don't expect this "bug" to be fixed.  I called it a bug, and I would have 
designed the API different nowadays, but in the past, I also wondered about 
this "magic" (0.5,0.5) shift.

But note that in the end, having integer coordinates denote pixel centers will 
usually lead to *less* magic shifts being needed than before.  The patch 
discussed here is a good example.

-- 
Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20071202/1edd6612/attachment.sig>


More information about the kde-core-devel mailing list