Ugly bug with TikZ in recent ConTeXt MKIV
Dear Hans, The following example worked perfectly fine with the version of ConTeXt MKIV I was using one hour ago (no idea which version, but something less than two weeks old). It still works OK with MKII and it works much worse in the real document that I'm using. In my document both images are completely screwed up. Here it's only the second one and if I comment out the first one, the second figure comes out fine, so I'm a bit clueless about what magic is going on inbetween. Nothing in TikZ has changed (I'm still using the "frozen" version from November 2009 from minimals), so it must have something to do with ConTeXt MKIV. It just messes up with some boxes on the page in some unpredictable way. \usemodule[tikz] \starttext \placefigure[force][BCl3 sigma]{x}{ \hbox to .9\textwidth{\hss\starttikzpicture \node[circle,inner sep=0pt] (B) at (0,0) {B}; \node[circle,inner sep=0pt] (a1) at ( 90:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a2) at (210:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a3) at (330:0.9cm) {Cl}; \draw (B)--(a1); \draw (B)--(a2); \draw (B)--(a3); \stoptikzpicture \hbox to 2cm{} \starttikzpicture[>=stealth,very thick] % \node[label=above:$a_1$] (a1) at ( 90:1cm) {$a_1$}; \node[circle,inner sep=0pt] (a1) at ( 90:1.2cm) {$a_1$}; \node[circle,inner sep=0pt] (a2) at (210:1.2cm) {$a_2$}; \node[circle,inner sep=0pt] (a3) at (330:1.2cm) {$a_3$}; % \draw[->] (0,0)--(-30:1cm); % \draw[->] (0,0)--( 90:1cm); % \draw[->] (0,0)--(210:1cm); \draw[->] (0,0)--(a1); \draw[->] (0,0)--(a2); \draw[->] (0,0)--(a3); \stoptikzpicture\hss} } \stoptext Thanks, Mojca
On Thursday 18 November 2010 00:00:33 Mojca Miklavec wrote:
Dear Hans,
The following example worked perfectly fine with the version of ConTeXt MKIV I was using one hour ago (no idea which version, but something less than two weeks old). It still works OK with MKII and it works much worse in the real document that I'm using.
In my document both images are completely screwed up. Here it's only the second one and if I comment out the first one, the second figure comes out fine, so I'm a bit clueless about what magic is going on inbetween.
Nothing in TikZ has changed (I'm still using the "frozen" version from November 2009 from minimals), so it must have something to do with ConTeXt MKIV. It just messes up with some boxes on the page in some unpredictable way.
\usemodule[tikz] \starttext \placefigure[force][BCl3 sigma]{x}{ \hbox to .9\textwidth{\hss\starttikzpicture \node[circle,inner sep=0pt] (B) at (0,0) {B}; \node[circle,inner sep=0pt] (a1) at ( 90:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a2) at (210:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a3) at (330:0.9cm) {Cl}; \draw (B)--(a1); \draw (B)--(a2); \draw (B)--(a3); \stoptikzpicture \hbox to 2cm{} \starttikzpicture[>=stealth,very thick] % \node[label=above:$a_1$] (a1) at ( 90:1cm) {$a_1$}; \node[circle,inner sep=0pt] (a1) at ( 90:1.2cm) {$a_1$}; \node[circle,inner sep=0pt] (a2) at (210:1.2cm) {$a_2$}; \node[circle,inner sep=0pt] (a3) at (330:1.2cm) {$a_3$}; % \draw[->] (0,0)--(-30:1cm); % \draw[->] (0,0)--( 90:1cm); % \draw[->] (0,0)--(210:1cm); \draw[->] (0,0)--(a1); \draw[->] (0,0)--(a2); \draw[->] (0,0)--(a3); \stoptikzpicture\hss} } \stoptext
Thanks, Mojca
Seems to work OK for me. Alan P.S. You can try (I did not look into "stealth"): \usemodule[tikz] \starttext \placefigure[force][BCl3 sigma]{x}{ \hbox to .9\textwidth{\hss\starttikzpicture \node[circle,inner sep=0pt] (B) at (0,0) {B}; \node[circle,inner sep=0pt] (a1) at ( 90:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a2) at (210:0.9cm) {Cl}; \node[circle,inner sep=0pt] (a3) at (330:0.9cm) {Cl}; \draw (B)--(a1); \draw (B)--(a2); \draw (B)--(a3); \stoptikzpicture \hbox to 2cm{} \starttikzpicture[>=stealth,very thick] \node[circle,inner sep=0pt] (a1) at ( 90:1.2cm) {$a_1$}; \node[circle,inner sep=0pt] (a2) at (210:1.2cm) {$a_2$}; \node[circle,inner sep=0pt] (a3) at (330:1.2cm) {$a_3$}; \draw[->] (0,0)--(a1); \draw[->] (0,0)--(a2); \draw[->] (0,0)--(a3); \stoptikzpicture\hss} } \placefigure [force] [BCl3 sigma bis]{x}{ \framed [frame=off,width=.9\textwidth] { \startMPcode label(textext("B"),(0,0)); for i=1 upto 3: pair p; p := (0.9cm,0) rotated (90+120(i-1)); draw (p scaled .25)--(p scaled .75); label(textext("Cl"),p); endfor \stopMPcode \hbox to 2cm{} \startMPcode for i=1 upto 3: pair p; p := (1.2cm,0) rotated (90+120(i-1)); drawarrow (0,0)--(p scaled .75) withpen pencircle scaled 1pt ;; label(textext("$a_"&decimal i&"$"),p); endfor \stopMPcode } } \stoptext
On 18-11-2010 12:00, Mojca Miklavec wrote:
Dear Hans,
The following example worked perfectly fine with the version of ConTeXt MKIV I was using one hour ago (no idea which version, but something less than two weeks old). It still works OK with MKII and it works much worse in the real document that I'm using.
In my document both images are completely screwed up. Here it's only the second one and if I comment out the first one, the second figure comes out fine, so I'm a bit clueless about what magic is going on inbetween.
Nothing in TikZ has changed (I'm still using the "frozen" version from November 2009 from minimals), so it must have something to do with ConTeXt MKIV. It just messes up with some boxes on the page in some unpredictable way.
I have no clue ... maybe some magic tikz variable (or shared register) that does not get reset? \usemodule[tikz] \starttext \startbuffer[1] \starttikzpicture \draw[->] (0,0)--(1cm,1cm); \stoptikzpicture \stopbuffer % \ruledhbox{\getbuffer[1]} \ruledhbox {\ruledhbox{\getbuffer[1]} \ruledhbox{\getbuffer[1]}} \ruledhbox {\ruledhbox{\getbuffer[1]}% \ruledhbox{\getbuffer[1]}} \ruledhbox{\getbuffer[1]} \stoptext Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
On Thu, Nov 18, 2010 at 09:48, Hans Hagen wrote:
On 18-11-2010 12:00, Mojca Miklavec wrote:
Dear Hans,
The following example worked perfectly fine with the version of ConTeXt MKIV I was using one hour ago (no idea which version, but something less than two weeks old). It still works OK with MKII and it works much worse in the real document that I'm using.
In my document both images are completely screwed up. Here it's only the second one and if I comment out the first one, the second figure comes out fine, so I'm a bit clueless about what magic is going on inbetween.
Nothing in TikZ has changed (I'm still using the "frozen" version from November 2009 from minimals), so it must have something to do with ConTeXt MKIV. It just messes up with some boxes on the page in some unpredictable way.
Hans, thanks a lot for the really nice simple example.
I have no clue ... maybe some magic tikz variable (or shared register) that does not get reset?
I have no clue either, but this has never been a problem before and it still works perfectly in both pdfTeX and XeTeX. So something really looks suspicious on the MKIV part. I would love to help debugging, but I have zero clue what and where to look for.
\usemodule[tikz]
\starttext
\startbuffer[1] \starttikzpicture \draw[->] (0,0)--(1cm,1cm); \stoptikzpicture \stopbuffer
% \ruledhbox{\getbuffer[1]}
\ruledhbox {\ruledhbox{\getbuffer[1]} \ruledhbox{\getbuffer[1]}}
\ruledhbox {\ruledhbox{\getbuffer[1]}% \ruledhbox{\getbuffer[1]}}
\ruledhbox{\getbuffer[1]}
\stoptext
Thanks, Mojca
On 18-11-2010 10:03, Mojca Miklavec wrote:
On Thu, Nov 18, 2010 at 09:48, Hans Hagen wrote:
On 18-11-2010 12:00, Mojca Miklavec wrote:
Dear Hans,
The following example worked perfectly fine with the version of ConTeXt MKIV I was using one hour ago (no idea which version, but something less than two weeks old). It still works OK with MKII and it works much worse in the real document that I'm using.
In my document both images are completely screwed up. Here it's only the second one and if I comment out the first one, the second figure comes out fine, so I'm a bit clueless about what magic is going on inbetween.
Nothing in TikZ has changed (I'm still using the "frozen" version from November 2009 from minimals), so it must have something to do with ConTeXt MKIV. It just messes up with some boxes on the page in some unpredictable way.
Hans,
thanks a lot for the really nice simple example.
I have no clue ... maybe some magic tikz variable (or shared register) that does not get reset?
I have no clue either, but this has never been a problem before and it still works perfectly in both pdfTeX and XeTeX. So something really looks suspicious on the MKIV part.
After some tracing and hard thinking Taco and I found interesting spaces showing up in the output that result from a kludge in tikz: Normally one will use \ignorespaces to catch spurious spaces but tikz invokes \nullfont assuming that its fontdimen 2 is zero. But, in context it no longer is (prelude to delayed font loading). You can put this in cont-new.mkiv as a temporary hack \let\normalnullfont\nullfont \def\nullfont {\fontdimen1\normalnullfont\zeropoint \fontdimen2\normalnullfont\zeropoint \fontdimen3\normalnullfont\zeropoint \fontdimen4\normalnullfont\zeropoint \fontdimen5\normalnullfont\zeropoint \fontdimen6\normalnullfont\zeropoint \fontdimen7\normalnullfont\zeropoint \normalnullfont} ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
2010/11/18 Hans Hagen
After some tracing and hard thinking Taco and I found interesting spaces showing up in the output that result from a kludge in tikz:
Normally one will use \ignorespaces to catch spurious spaces but tikz invokes \nullfont assuming that its fontdimen 2 is zero. But, in context it no longer is (prelude to delayed font loading).
You can put this in cont-new.mkiv as a temporary hack
\let\normalnullfont\nullfont
\def\nullfont {\fontdimen1\normalnullfont\zeropoint \fontdimen2\normalnullfont\zeropoint \fontdimen3\normalnullfont\zeropoint \fontdimen4\normalnullfont\zeropoint \fontdimen5\normalnullfont\zeropoint \fontdimen6\normalnullfont\zeropoint \fontdimen7\normalnullfont\zeropoint \normalnullfont}
Perhaps this should be forwared to Christian Feuersaenger. He is very open to bug reports and will probably be willing to fix it in some way in TikZ. Regards, -- Vedran Miletić
On 18-11-2010 5:56, Vedran Miletić wrote:
2010/11/18 Hans Hagen
After some tracing and hard thinking Taco and I found interesting spaces showing up in the output that result from a kludge in tikz:
Normally one will use \ignorespaces to catch spurious spaces but tikz invokes \nullfont assuming that its fontdimen 2 is zero. But, in context it no longer is (prelude to delayed font loading).
You can put this in cont-new.mkiv as a temporary hack
\let\normalnullfont\nullfont
\def\nullfont {\fontdimen1\normalnullfont\zeropoint \fontdimen2\normalnullfont\zeropoint \fontdimen3\normalnullfont\zeropoint \fontdimen4\normalnullfont\zeropoint \fontdimen5\normalnullfont\zeropoint \fontdimen6\normalnullfont\zeropoint \fontdimen7\normalnullfont\zeropoint \normalnullfont}
Perhaps this should be forwared to Christian Feuersaenger. He is very open to bug reports and will probably be willing to fix it in some way in TikZ.
feel free to do so; probably something like this is needed: \def\pushnullfont {\edef\popnullfont {\fontdimen2\nullfont\the\fontdimen2\nullfont \fontdimen3\nullfont\the\fontdimen3\nullfont \fontdimen4\nullfont\the\fontdimen4\nullfont}% \fontdimen2\nullfont 0pt\relax \fontdimen3\nullfont 0pt\relax \fontdimen4\nullfont 0pt\relax} \pushnullfont .... \popnullfont as fontdimens are assigned global. Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
Mojca, Can you test the beta on the ftp server and see if tikz works better? (I hate this kind of patching as it is a work around assumptions that are somewhat wrong, in this case the fact that nullfont has zero parameters, and it makes the mkiv code messier.) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
participants (4)
-
Alan BRASLAU -
Hans Hagen -
Mojca Miklavec -
Vedran Miletić