Possible bug with paths and strokes

["John C. Turnbull" <oz...@ozemail.com.au>]

The attached SVG renders correctly in ASV and Opera but with Batik the
stroke path is angular and does not follow the actual curve of the path as I
believe it should.

 

Is this a bug?  I am using the latest SVN source.

 

Thanks,

 

John

Re: Possible bug with paths and strokes

[Helder Magalhães <he...@gmail.com>]

>> This might be related to issue 41294 [1]
> Should I file it as a bug or do you feel it's just a duplicate of 41294?

Well, I was mostly trying to narrow the issue a bit so Batik/Java 2D
experts could jump in with some research already made. :-)

If no one has the chance to provide feedback in a couple of days, in
my opinion this should be filed separately as I have a feeling that
this particular test case is triggering a new Batik issue... :-|

I'm not sure if this can also be related to "lack" of coordinate
precision: Floats [1] are used when parsing path "curve" command
coordinates, and given that the coordinates are rather tiny (combined
with the involved transforms and initial viewport), this could be a
source of erroneous behavior...

Regards,

 Helder Magalhães

[1] http://svn.apache.org/viewvc/xmlgraphics/batik/trunk/sources/org/apache/batik/parser/PathParser.java?view=annotate#l318

---------------------------------------------------------------------
To unsubscribe, e-mail: batik-users-unsubscribe@xmlgraphics.apache.org
For additional commands, e-mail: batik-users-help@xmlgraphics.apache.org

RE: Possible bug with paths and strokes

["John C. Turnbull" <oz...@ozemail.com.au>]

Thanks Helder,

Should I file it as a bug or do you feel it's just a duplicate of 41294?

John

> -----Original Message-----
> From: Helder Magalhães [mailto:helder.magalhaes@gmail.com]
> Sent: Wednesday, 21 January 2009 22:14
> To: batik-users@xmlgraphics.apache.org
> Subject: Re: Possible bug with paths and strokes
> 
> > The attached SVG renders correctly in ASV and Opera
> 
> Firefox (at least 3.0.x) also seems to render it as expected.
> 
> 
> > Is this a bug?  I am using the latest SVN source.
> 
> Confirmed with latest SVN source and releases 1.6 and 1.7 (checking for
> a possible regression!).
> 
> This might be related to issue 41294 [1] which points out towards
> issues in the Java 2D renderer, given the coordinate values (rather
> small). Java environment in my tests was (Sun's) jdk1.6.0_11. Note that
> it seems that there were some recent improvements [2] in Sun's
> renderer, although they don't seem to fit to this case (I was able to
> reproduce using update 11, while the change refers improvements in
> update 10).
> 
> Some efforts were made to reduce the problem:
>  * Changed/removed attributes but the behavior didn't change;
>  * Made the height and width set to 100% so that the document size
> follows the canvas size (to check for particular coordinate precision
> issues).
> Although nothing substantial was revealed, a slightly shorter test case
> was obtained (attached).
> 
> 
> > Thanks,
> > John
> 
> Hope this helps,
> 
>   Helder
> 
> 
> [1] https://issues.apache.org/bugzilla/show_bug.cgi?id=41294
> [2] https://issues.apache.org/bugzilla/show_bug.cgi?id=43411#c11


---------------------------------------------------------------------
To unsubscribe, e-mail: batik-users-unsubscribe@xmlgraphics.apache.org
For additional commands, e-mail: batik-users-help@xmlgraphics.apache.org

Re: Possible bug with paths and strokes

[Helder Magalhães <he...@gmail.com>]

> The attached SVG renders correctly in ASV and Opera

Firefox (at least 3.0.x) also seems to render it as expected.


> Is this a bug?  I am using the latest SVN source.

Confirmed with latest SVN source and releases 1.6 and 1.7 (checking
for a possible regression!).

This might be related to issue 41294 [1] which points out towards
issues in the Java 2D renderer, given the coordinate values (rather
small). Java environment in my tests was (Sun's) jdk1.6.0_11. Note
that it seems that there were some recent improvements [2] in Sun's
renderer, although they don't seem to fit to this case (I was able to
reproduce using update 11, while the change refers improvements in
update 10).

Some efforts were made to reduce the problem:
 * Changed/removed attributes but the behavior didn't change;
 * Made the height and width set to 100% so that the document size
follows the canvas size (to check for particular coordinate precision
issues).
Although nothing substantial was revealed, a slightly shorter test
case was obtained (attached).


> Thanks,
> John

Hope this helps,

  Helder


[1] https://issues.apache.org/bugzilla/show_bug.cgi?id=41294
[2] https://issues.apache.org/bugzilla/show_bug.cgi?id=43411#c11