[Poppler-bugs] [Bug 49523] Splash::arbitraryTransformImage causes bogus memory allocation size

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=49523

thomasf <Thomas.Freitag at alfa.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #61073|0                           |1
        is obsolete|                            |

--- Comment #5 from thomasf <Thomas.Freitag at alfa.de> 2012-05-12 00:34:12 PDT ---
Created attachment 61476
  --> https://bugs.freedesktop.org/attachment.cgi?id=61476
Avoid bogus memory allocation size

You're really great in detection of regressions: I must admit that when I had a
first look at the outputs today morning I hadn't any clue what You mean. Only
when I encountered in the debugger that it is not only one image but that it is
striped I saw what You meant.
Then I first tried to remove my changes of xpdf code, but the result wasn't
even better. With the new approach now I just use the calculation of th to
detect if the image is rotated, and if so, I double the scaled height to ensure
that we have enough pixels. This solves the problem why I introduced it
(Algorithmics - The Spirit of Computing, 3rd Ed.pdf, page 284 and 288, at least
when rendering it with 150 dpi we get an acceptable quality), it still renders
the PDF from Ralph and because the factor is now always the same we again get
an acceptable quality in bug176320.pdf.
I started my regtest, but I don't want to wait until it finishes because I have
to leave (tennis season has started :-) ), perhaps You'll find the time to
regtest it, too, You're much better in finding regressions, and we can discuss
it tomorrow afternoon.

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.