From: Andrew R. <and...@us...> - 2007-12-01 08:58:50
|
On Fri, Nov 30, 2007 at 05:35:06PM -0800, Alan Irwin wrote: > On 2007-11-29 16:04-0800 Alan W. Irwin wrote: > > >Meanwhile, I will only bring up the xyshow issue on the cairo list. > > When I did more thorough homework on this xyshow issue including looking at > the libcairo code, it eventually boiled down to a one-line cairo patch which > I submitted to the cairo list. It turned out their PostScript backend guy > Adrian Johnson had independently come up with exactly the same one-liner and > committed it just a few hours before my post! (See > http://lists.cairographics.org/archives/cairo/2007-November/012152.html). > > It should take a while to propagate this patch to distro releases, but I > have > tried to speed up this process for Debian (and eventually Ubuntu) with > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=453718 . Anyhow, except > for those propagation delays, this resolves the xyshow issue. Good news. > That leaves the remaining "/typecheck error for definefont" issue. All my > recent PLplot pscairo results have been for the pango-1.16.4/cairo-1.4.6 > stack that I built to mimic as closely as possible the results I had for my > debian oldstable=sarge system. I have now rebuilt PLplot and pscairo using > the much later pango-1.18.3/cairo-1.4.10 that you get by default for Debian > testing. > > For that latest version of PLplot, I get the "special" definefont pattern > as follows: > > software@raven> grep '>> definefont' test/*.pscairo > test/x07c.pscairo:>> definefont pop > test/x23c.pscairo:>> definefont pop > test/x23c.pscairo:>> definefont pop > test/x24c.pscairo:>> definefont pop > > where ">>" is the end of the dictionary (font) definition. For all those > examples, I also get the "/typecheck error for definefont" when I use the gv > --noquiet command. If I copy those examples to my oldstable box, the gv > application plots them without any showstopper errors. However, when > looking > at the 24th "peace-flag" example, the Korean version of "peace" is replaced > by blank. Meanwhile, on the debian testing side c/x24c -dev xcairo produces > a good peace flag except for the Korean version of "peace" which is about > 1/tenth size. This "Korean" problem also shows up for -dev pngcairo > > So my mental model of what is going on here is there is some issue with the > Korean fonts on my Debian testing system. For -dev xcairo and pngcairo the > issue shows up as a "peace" word which is much too small. -dev pscairo tries > to handle the problem with the special definefont pattern which has the > above signature, but makes some type error in the structure that is set up. > Modern gv/ghostscript errors out with a /typecheck error while older > gv/ghostscript just produces a blank for this case. > > Andrew, you have already confirmed part of this mental model when you showed > that results generated on my system with the above definefont signature > generated the "/typecheck error for definefont" error message for your > gutsy version of gv/ghostscript. > > I can well believe that Ubuntu has been a little more careful about font > issues than Debian so it makes sense that Andrew never sees the problem for > results generated by him on Ubuntu. However, Andrew, could you specifically > confirm that conclusion by looking for the special signature for definefont > that I found above? It should not be there for your Ubuntu results, but it > should be there for your Debian testing platform results, but I want to be > sure this is the case. Alan, I can confirm that the special definefont line ">> definefont" does not appear in my Ubuntu generated plots. I do not have access to the debian testing plots here at home to check them I'm afraid. I can confirm that I did not see any errors with gv --noquiet on these plots though. Since the cairo version we are using should be the same, then the only difference I can see is the installed fonts. Andrew |