[poppler] Annot border regressions

[Carlos Garcia Campos]

Carlos Garcia Campos <carlosgc at gnome.org> writes:

> I've updated the patches to not break the qt frontend, and I've run the
> complete test suite for all the backends. This time I've found a few
> differences.
>
> bug188109.pdf
>  - Page 60: now renders a border we didn't. Acroread renders the border
>  too.
>
> att_klein_wired.f4.pdf
>  - Page 29: we are now rendering a border in a link, acroread doesn't
>  even open the file.
>
> bug-poppler30580.pdf
>  - Contains several square annots without any border specified. We are
>  now using the default border in this case, but acroread doesn't. It's
>  because these annots have an XAP entry for the appearance stream. When
>  square annots have an AP we ignore the border, and use the provided AP
>  instead. Does anybody know what XAP is? Is it part of a PDF extension?
>  Should we parse those entries the same way we do for AP?
>
> bug102145.pdf
> damino-oss-lang-de.pdf
>  - URL links should have a border that we didn't render.
>
> links/817-6223.pdf
>  - This is funny. It contains white links without any border, so now
>  we are drawing the default border, that seems to be invisible in most
>  of the cases, but in some cases it intersects with the link text, so
>  that it looks like if the text was not correctly rendered. So, at a
>  first glance it looks like a regression, but it's actually a buggy
>  document. Acroread renders the same. White links sounds like a
>  workaround for pdf viewers drawing borders in links by default
>  (following the spec)
>
> There are a few files that didn't render anything where we now seem to
> draw borders at random places, but they don't even open in acroread so
> it's difficult to say.
>
> There are also several docs containing check and radio form field buttons
> that we are now render differently, and I would say worse, so I'll look
> at those. Apart from this the only actual regression I see is the square
> annots with an XAP entry of bug-poppler30580.pdf.

I've just pushed the patches. The only regression is with
bug-poppler30580.pdf that I'll fix in a follow up patch once Leonard
clarifies why we behave differently in that case.

Regards, 
-- 
Carlos Garcia Campos
PGP key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x523E6462
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/poppler/attachments/20131205/4bfe5ef1/attachment.pgp>