<html> <head> <base href="https://bugs.freedesktop.org/" /> </head> <body> <div> <a class="bz_bug_link bz_status_NEW " title="NEW --- - Image renders as black blob" href="https://bugs.freedesktop.org/show_bug.cgi?id=61719#c12">Comment # 12</a> on <a class="bz_bug_link bz_status_NEW " title="NEW --- - Image renders as black blob" href="https://bugs.freedesktop.org/show_bug.cgi?id=61719">bug 61719</a> from <a class="email" href="mailto:tsdgeos@terra.es" title="Albert Astals Cid <tsdgeos@terra.es>"> Albert Astals Cid</a> <pre>(In reply to <a href="show_bug.cgi?id=61719#c11">comment #11</a>) > (In reply to <a href="show_bug.cgi?id=61719#c10">comment #10</a>) > > Thomas two things i was wondering, have you run this on "default" mode and > > seen how much stuff it changes? > > Hmmh, I started a test with the wrong implementation (looking at the line > width), but I broke it after 150 tests because 30% of tests failed. It's now > probably better, because I reduced the limit from less than 1 pixel to less > than 0.2 pixel and look correctly at the width or height of the clipping > area. > BUT: The problem here is that a PDF is converted to EPS, loosing it tiling > pattern, and then converted back to PDF. So it's not a "normal" PDF, and > since I don't get a "normal" PDF with such a problem I don't like to change > the accepted rendering behaviour. > And because the "pattern" are thin lines, I think that to choose "Enhance > thin lines" is the right switch (even if somewhat misused, but I don't want > to increase the number of switches every time we get a new, not general > solvable problem) Oki > > > > > Also, do you think that instead of calculating xMinFP all the time we create > > a SplashXPath maybe we could do it just when we're going to use it? > > In the constructor of XPath the path elements are already adjusted according > to the hints, so the result is no more usable for our case. So we have to > run once again over the path points and transform them to get the bounding > box. But if You think that it is too expensive to do it always in the > constructor and that we need it nowhere else in futur, of course it is > doable (and at least we will only do it if we have a "single point width or > height rectangle"). Just give a go if You accept the output behaviour but > want to have it only if definitely needed and I'll implement a helper > function in Splash. Yeah given we only use it in very specific cases (i.e. we have thin line mode enabled) i would prefer if we move the code to when it's really needed not making the general case slower for "no reason"</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the assignee for the bug.</li> </ul> </body> </html>