<?xml version="1.0" encoding="UTF-8" standalone="yes"?><oembed><version><![CDATA[1.0]]></version><provider_name><![CDATA[Software is Crap]]></provider_name><provider_url><![CDATA[https://davmac.wordpress.com]]></provider_url><author_name><![CDATA[davmac]]></author_name><author_url><![CDATA[https://davmac.wordpress.com/author/davmac/]]></author_url><title><![CDATA[Cairo 1.12.0 &#8211; buggy as&nbsp;buggery?]]></title><type><![CDATA[link]]></type><html><![CDATA[<p><a href="http://cairographics.org/news/cairo-1.12.0/">Cairo 1.12.0 has been released</a> with a swag-full of new features (such as XCB backend) and improvements. Apparently. For me, it causes random text corruption, noticeable most prominently in Firefox (when built with &#8211;enable-system-cairo of course) but also in a few other GTK-based apps. I&#8217;ve tried compiling with both GCC 4.7.0, 4.6.3 and LLVM 3.0 without any resolution. Cairo 1.10.2 doesn&#8217;t produce corruption, 1.12.0 does. For the moment I recommend holding off on the upgrade.</p>
<p>The situation would be a lot better if Cairo had a halfway decent test suite. It does have <em>a</em> test suite, but when I run it (on 1.12.0 or 1.10.2) I get a huge number of failures which are completely inexplicable to me and <a href="http://www.linuxfromscratch.org/blfs/view/svn/x/cairo.html">apparently to LFS authors also</a>. Quotes from that page:</p>
<blockquote><p>As the test suite is currently unreliable, it is best to simply skip it at this time.</p></blockquote>
<p>and</p>
<blockquote><p>Note that the tests take a long time to run and many of them fail for unknown reasons.</p></blockquote>
<p>&#8230; which pretty much sums up my experience as well. What a joke.</p>
<p><strong>Update:</strong> <a href="https://bugs.freedesktop.org/show_bug.cgi?id=47266">seems I&#8217;m not the only one who&#8217;s seen this.</a> Not clear at this stage whether it&#8217;s really a bug in cairo, in the radeon driver, or the EXA extension.</p>
]]></html></oembed>