W3C home > Mailing lists > Public > www-svg@w3.org > October 2011

Re: patternTransform -- and toroidal wraps

From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
Date: Tue, 4 Oct 2011 16:49:58 +0200
To: www-svg@w3.org
Message-Id: <201110041649.58216.Dr.O.Hoffmann@gmx.de>
Some time ago in my tutorial I wrote an article about
those wallpaper groups and additionally on penrose
tiles and it did not take a long time to find out that
due to different bugs in viewers, the pattern feature
was not ready to use with typical viewers.
Typical reasons are blurring of the pattern for example in 
Opera, scaling/transform problems in firefox (not sure, 
if already fixed in current the version), some funny gaps 
for example in the adobe plugin and maybe firefox (?)
and of course the obvious problem with overflow.
I think, really undefined is only the behaviour, if there
is an overlap. This was discussed in this mailing list,
but there was no agreement to define this for SVG 1.1
second edition, therefore I think it was added a note,
that the behaviour is not defined - from my point of
view this means, authors have only take into account
differences, if there is an overlap, else the undefinedness
will have no visible effect, if the overflow property is not
always ignored as currently. To interprete overflow="visible",
as overflow="hidden" is obviously always a bug of the viewer,
if there is something outside of the viewBox. Only if the
behaviour for overlapping painted elements is surprising, this
is no bug, because behaviour is undefined. For those cases,
authors still have to clip their master template in such a way, that 
overlaps are avoided. For some symmetry groups it is not meaningful
to clip to a rectangle, but often it is no problem to clip to another
shape - therefore a combination of overflow="visible" with some
author defined clipping path can already help to solve many
practical problems like the accuracy gaps of some viewers.

However, my conclusion from the previous discussion and the
observation of viewer bugs was, that SVG 1.1 pattern and
its implementations are not really ready for use for advanced
pattern. To get a defined behaviour, authors still have to
do this for periodic pattern in the same way as for aperiodic
pattern as penrose tiles - with use and transform ;o)  
Especially p3... and p6... groups are always a problem, because
one practically has always problems with overflows and 
artifacts like accuracy gaps etc. For groups with lower symmetries
or symmetries more related to a rectangle one can more often work 
around the problems with some clever considerations to get some 
meaningful result with pattern.
And of course, over the years, more and more authors will come
to similar conclusions about viewer behaviour and will ask again 
about the issue ;o)


Olaf
Received on Tuesday, 4 October 2011 14:50:27 GMT

This archive was generated by hypermail 2.3.1 : Friday, 8 March 2013 15:54:49 GMT