Xlib issue
Daniel Price <[email protected]>
| Newsgroups | gmane.comp.lib.cairo |
|---|---|
| Message-ID | <[email protected]> |
Dear cairo developers, We've recently developed a 2D scientific plotting library called giza (because it's based on cairo!), meant as a low-dependency backend replacement for PGPLOT in Astronomy software (see http://giza.sourceforge.net/). There is one nasty bug that I've been trying to chase in the Xlib driver that seems to be a cairo thing, but seemingly only in older versions of cairo. It occurs when I open a cairo-xlib surface for the second time (after having closed the first one) and consists of errors like: =========================================== X Error of failed request: 150 Major opcode of failed request: 144 (RENDER) Minor opcode of failed request: 23 (RenderCompositeGlyphs8) Serial number of failed request: 53 Current serial number in output stream: 3998 =========================================== I have found references to similar things in earlier threads w.r.t multi-threaded applications. For example this thread: http://lists.cairographics.org/archives/cairo/2009-December/018629.html giza doesn't use multiple threads but maybe opening a second xlib device has a similar effect? It seems like something is not being closed off properly when closing the first xlib surface, but I can't find anything in my code that could cause that (I've boiled it down to a reasonably simple test case). I should point out that the context is completely destroyed and then recreated before opening the device for the second time. My questions for the list are: a) Is/was this was a known issue with cairo-xlib devices? b) If so, what is the minimum version of cairo that it was fixed in? c) Is there a workaround to prevent this bug with older versions of cairo? (ideally I would prefer to be able to use whatever cairo is installed by default on the user's system, rather than enforcing installation of a new version) Any light that can be shed on this would be much appreciated. Thanks, Daniel -- cairo mailing list [email protected] http://lists.cairographics.org/mailman/listinfo/cairo