https://bugs.documentfoundation.org/show_bug.cgi?id=128879

--- Comment #11 from Dave Gilbert <[email protected]> ---
I've got some sympathy for the pdf code here...

Looking at the two PDF attachments, uncompressed and in meld, the biggest
differences are the newer one has a bunch of added bitmap/images embedded.

Digging into the example screwdriver in inkscape, there are two
paths which have hideously complex masking and filling on, and I think they do
have
a forward reference which I guess Xisco's patch made the reference work and now
the PDF code has tried to do masking.

In particular, inkscape says path 106 has mask g, fill i,
mask g has a reference to 
  fill: url#h (i.e. to the next fill - so not yet defined)

gradient h:
  has a graident transform and is linked back to a nice simple fill c


   <linearGradient id="c"><stop offset="0" stop-color="#fff"/><stop
offset="1"/></linearGradient>

<mask id="g" height="50.138" maskUnits="userSpaceOnUse" width="45.542"
x="20.098" y="17.939"><g filter="url(#a)">
  <path d="m30.168 19.316-10.438 9.329-5.914-6.618 10.439-9.328z"
fill="url(#h)"/></g></mask>

<linearGradient id="h" gradientTransform="matrix(.7457 -.6663 .6663 .7457
4.3633 147.9164)" gradientUnits="userSpaceOnUse" x1="97.9258" x2="97.9258"
xlink:href="#c" y1="-80.3257" y2="-83.1696"/>

so I guess it's something to do with the pdf exporters masking code or it's
handling those complex fills.  Perhaps it just gave up and tried to spit out
the image.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to