Problem with MkIV and TikZ picture remembering

newer
\placefootnotes at the end of each...

older
Re: [NTG-context] component not...

Jörg Weger

3 Mar 2015 3 Mar '15

8:23 a.m.

In the minimal working example below inside of a text two words should be printed inside of rounded TikZ rectangles and should be connected by an TikZ arrow pointing from the first word to the other. TikZ offers the remember picture/overlay option for that. The MWE works in MkII (command “texexec”) which I am not using, but not in the latest MkIV (command “context”). The problem might be that according to the PGF/TikZ manual “You need to use a driver that supports picture remembering and you need to run TEX twice” (16.13.1 Referencing a Node in a Different Picture). So my questions are: 1.) Does LuaTeX support TikZ picture remembering? 2.) If so, how can I get ConTeXt MkIV to “run TeX twice”? 3.) If all of the above worked, would that work also if the inline TikZ pictures to be remembered are on another file referenced via \input? I have seen a solution similar to what I want to achieve in the METAFUN manual but I don’t have a clue how to get that METAPOST alternative working. Greetings Jörg %%%%%%%%%%%%%%%% START MWE %%%%%%%%%%%%%%%%%%%%%%%%%%%%% \usemodule[tikz] \usetikzlibrary[shapes.misc,arrows] \starttext \tikz[baseline,remember picture] \node[rounded rectangle,draw,anchor=text] (n1) {this}; belongs to \tikz[baseline,remember picture] \node[rounded rectangle,draw,anchor=text] (n2) {that}; \starttikzpicture [remember picture,overlay] \draw[->] (n1.north) to [bend left] (n2.north); \stoptikzpicture \stoptext %%%%%%%%%%%%%%%% STOP MWE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Show replies by date

Jörg Weger

3 Mar 3 Mar

8:25 a.m.

Wolfgang Schuster

8:54 a.m.

...

Am 03.03.2015 um 08:25 schrieb Jörg Weger :

In the minimal working example below inside of a text two words should be printed inside of rounded TikZ rectangles and should be connected by an TikZ arrow pointing from the first word to the other.

TikZ offers the remember picture/overlay option for that.

The MWE works in MkII (command “texexec”) which I am not using, but not in the latest MkIV (command “context”).

The problem might be that according to the PGF/TikZ manual “You need to use a driver that supports picture remembering and you need to run TEX twice” (16.13.1 Referencing a Node in a Different Picture).

So my questions are:

1.) Does LuaTeX support TikZ picture remembering?

2.) If so, how can I get ConTeXt MkIV to “run TeX twice”?

It depends, when you have a command which writes something into contexts own auxiliary file context will make another run but when you have a module like tikz which uses its own auxiliary file just make another run with „context <filename>”.

...

3.) If all of the above worked, would that work also if the inline TikZ pictures to be remembered are on another file referenced via \input?

The problem here is the content of tikz auxiliary file, when I process your document with MkII it contains \pgfsyspdfmark {pgfid1}{5006382}{46179011} \pgfsyspdfmark {pgfid2}{10902017}{46179011} \pgfsyspdfmark {pgfid3}{4661756}{45230922} but with MkIV i get the following result: \gdef \pgf@sys@pdf@mark@pos@pgfid1 {\pgf@x =0sp\pgf@y =0sp} \gdef \pgf@sys@pdf@mark@pos@pgfid2 {\pgf@x =0sp\pgf@y =0sp} \gdef \pgf@sys@pdf@mark@pos@pgfid3 {\pgf@x =0sp\pgf@y =0sp} There is something wrong with the way tikz saves the coordinates of the nodes in MkIV. Wolfgang

Ulrike Fischer

1:38 p.m.

Am Tue, 3 Mar 2015 08:54:00 +0100 schrieb Wolfgang Schuster: I wanted to try the example in my context standalone installation and got an "undefined command error" on the \usetikzlibrary. After a bit digging I could repair this with "first-install.bat --modules="t-tikz" which seems to get me the newest tikz. But during this digging I saw -- that tikz is not mentioned on the module list: http://modules.contextgarden.net/ -- Following the link to the module (http://modules.contextgarden.net/TikZ) from http://wiki.contextgarden.net/TikZ gives an "internal error message" -- the installation advices on contextgarden doesn't explain what to do if you want to update modules too. I'm always using (on windows) first-setup.bat --keep as explained here http://wiki.contextgarden.net/ConTeXt_Standalone#Updating_2. But this doesn't seem to update the modules. Is first-setup.bat --all better if one simply wants the newest complete standalone? -- Ulrike Fischer http://www.troubleshooting-tex.de/

Mojca Miklavec

2:11 p.m.

On Tue, Mar 3, 2015 at 1:38 PM, Ulrike Fischer wrote:

...

I wanted to try the example in my context standalone installation and got an "undefined command error" on the \usetikzlibrary. After a bit digging I could repair this with "first-install.bat --modules="t-tikz" which seems to get me the newest tikz.

But during this digging I saw

-- that tikz is not mentioned on the module list: http://modules.contextgarden.net/

That site is used for uploading modules by authors. I never thought about users trying to learn about existence of modules via that site.

...

-- Following the link to the module (http://modules.contextgarden.net/TikZ) from http://wiki.contextgarden.net/TikZ gives an "internal error message"

I removed the link.

...

-- the installation advices on contextgarden doesn't explain what to do if you want to update modules too.

Do you want to say that the module gets removed next time when you run first-setup.bat? If that is the case, it should be fixed. Updating works for me and also keeps the modules up to date.

...

I'm always using (on windows) first-setup.bat --keep

I would advise you against using the --keep option unless there are some other/strong reasons to do so. But you shouldn't need keep just for the sake of a potential bug. I would need to check again what exactly --all does. But --modules=all should install all modules and --modules=t-tikz should be sufficient to keep that module. However I don't know what happens if you use --modules=t-somethingelse next time. Maybe that removes the modules that you installed previously as it expects the complete list of modules. I need to check. Mojca

Ulrike Fischer

2:27 p.m.

Am Tue, 3 Mar 2015 14:11:56 +0100 schrieb Mojca Miklavec:

...

...
-- the installation advices on contextgarden doesn't explain what to do if you want to update modules too.

...

Do you want to say that the module gets removed next time when you run first-setup.bat? If that is the case, it should be fixed. Updating works for me and also keeps the modules up to date.

I have no idea ;-).

...

...
I'm always using (on windows) first-setup.bat --keep

I would advise you against using the --keep option unless there are some other/strong reasons to do so.

Then the instructions here are quite misleading: http://wiki.contextgarden.net/ConTeXt_Standalone#Updating_2 I'm not a regularly context user, I maintain the installation mostly for tests and want "an up-to-date, complete (with all modules)" installation and was looking for a command that compares to "tlmgr update --all" from texlive. Did I got it right now that I should do "first-setup.bat" and that "first-setup.bat --keep" doesn't update the modules? -- Ulrike Fischer http://www.troubleshooting-tex.de/

Mojca Miklavec

4:29 p.m.

On Tue, Mar 3, 2015 at 2:27 PM, Ulrike Fischer wrote:

...

...
...
I'm always using (on windows) first-setup.bat --keep

I would advise you against using the --keep option unless there are some other/strong reasons to do so.

Then the instructions here are quite misleading: http://wiki.contextgarden.net/ConTeXt_Standalone#Updating_2

Thank you. Removed. I need to review the page. There are some weird statements there. If you are familiar with rsync: all that --keep does is omit the "--delete" flag in rsync modules. It's not exactly the same, but it is roughly "equally evil" as setting the TeX Live's option to keep the removed modules installed. Except that here it would accumulate all the files that ever existed in ConTeXt since the first day when the user first installed the distribution. When, say, fonts in TeX Live move around, users would end up with duplicate files at two locations.

...

I'm not a regularly context user, I maintain the installation mostly for tests and want "an up-to-date, complete (with all modules)" installation and was looking for a command that compares to "tlmgr update --all" from texlive.

Did I got it right now that I should do "first-setup.bat" and that "first-setup.bat --keep" doesn't update the modules?

No, --keep isn't related to updating modules at all. The flag would help you keep things that you install in your tree from your own sources (something that doesn't come from the distribution). But it would also keep all other outdated files. It's waaaaaay better to keep your own files elsewhere. (Even if simply under "texmf-local" that's created for you. But not in "texmf-modules" to "texmf-context" or alike.) Mojca

Ulrike Fischer

2:51 p.m.

Am Tue, 3 Mar 2015 14:11:56 +0100 schrieb Mojca Miklavec:

...

...
-- that tikz is not mentioned on the module list: http://modules.contextgarden.net/

That site is used for uploading modules by authors. I never thought about users trying to learn about existence of modules via that site.

tikz isn't mentioned here too: http://wiki.contextgarden.net/Modules -- Ulrike Fischer http://www.troubleshooting-tex.de/

Aditya Mahajan

6:12 p.m.

On Tue, 3 Mar 2015, Jörg Weger wrote:

...

In the minimal working example below inside of a text two words should be printed inside of rounded TikZ rectangles and should be connected by an TikZ arrow pointing from the first word to the other.

TikZ offers the remember picture/overlay option for that.

I have seen a solution similar to what I want to achieve in the METAFUN manual but I don’t have a clue how to get that METAPOST alternative working.

Here is how to get the metafun alternative working. 1. I don't understand why I need to manually input the mp-abck file; that should be input automatically. 2. The metafun alternative is a bit inconvenient. Suppose you want to draw a bunch of such graphics (words connected by arrows). Then, you will need to ensure that node names are unique. (Is there a way to get around that?) 3. I did not bother to set the style of the boxes and arrows. So it looks a bit dry. \startMPdefinitions % I don't know why this does not get loaded automatically. input mp-abck.mpiv; \stopMPdefinitions \defineoverlay[highlight] [\positionoverlay{highlightoverlay}] \startpositionoverlay{highlightoverlay} \setMPpositiongraphic{n1}{highlightgraphic}{from=n1,to=n2} \stoppositionoverlay \startMPpositiongraphic{highlightgraphic} save n; path n[]; initialize_box(\MPpos{\MPvar{from}}); n[1] := pxy; initialize_box(\MPpos{\MPvar{to}}); n[2] := pxy; draw n[1]; draw n[2]; drawarrow (center n[1]){dir 45} .. {dir -45}(center n[2]) cutbefore n[1] cutafter n[2]; anchor_box(\MPanchor{\MPvar{from}}); \stopMPpositiongraphic \setuppapersize[A6] \setupbackgrounds[page][background=highlight] \starttext \hpos{n1}{this} belongs to \hpos{n2}{that} \stoptext Aditya

Hans Hagen

8:04 p.m.

On 3/3/2015 6:12 PM, Aditya Mahajan wrote:

...

On Tue, 3 Mar 2015, Jörg Weger wrote:

...
In the minimal working example below inside of a text two words should be printed inside of rounded TikZ rectangles and should be connected by an TikZ arrow pointing from the first word to the other.

TikZ offers the remember picture/overlay option for that.

I have seen a solution similar to what I want to achieve in the METAFUN manual but I don’t have a clue how to get that METAPOST alternative working.

Here is how to get the metafun alternative working.

1. I don't understand why I need to manually input the mp-abck file; that should be input automatically.

2. The metafun alternative is a bit inconvenient. Suppose you want to draw a bunch of such graphics (words connected by arrows). Then, you will need to ensure that node names are unique. (Is there a way to get around that?)

3. I did not bother to set the style of the boxes and arrows. So it looks a bit dry.

\startMPdefinitions % I don't know why this does not get loaded automatically. input mp-abck.mpiv; \stopMPdefinitions

\defineoverlay[highlight] [\positionoverlay{highlightoverlay}]

\startpositionoverlay{highlightoverlay} \setMPpositiongraphic{n1}{highlightgraphic}{from=n1,to=n2} \stoppositionoverlay

\startMPpositiongraphic{highlightgraphic} save n; path n[];

initialize_box(\MPpos{\MPvar{from}}); n[1] := pxy; initialize_box(\MPpos{\MPvar{to}}); n[2] := pxy;

draw n[1]; draw n[2];

drawarrow (center n[1]){dir 45} .. {dir -45}(center n[2]) cutbefore n[1] cutafter n[2];

anchor_box(\MPanchor{\MPvar{from}}); \stopMPpositiongraphic

\setuppapersize[A6]

\setupbackgrounds[page][background=highlight]

\starttext \hpos{n1}{this} belongs to \hpos{n2}{that} \stoptext

\startMPdefinitions % I don't know why this does not get loaded automatically. Answer: seldom needed. input mp-abck.mpiv; \stopMPdefinitions \defineoverlay[highlight] [\positionoverlay{highlightoverlay}] \startMPpositiongraphic{highlightgraphic} save n; path n[]; initialize_box(\MPpos{\MPvar{from}}); n[1] := pxy; initialize_box(\MPpos{\MPvar{to}}); n[2] := pxy; draw n[1]; draw n[2]; if ypart center n[1] > ypart center n[2] : drawarrow (center n[1]){dir 225} ... {dir -225}(center n[2]) cutbefore n[1] cutafter n[2] ; else : drawarrow (center n[1]){dir 45} ... {dir -45}(center n[2]) cutbefore n[1] cutafter n[2] ; fi ; anchor_box(\MPanchor{\MPvar{from}}); \stopMPpositiongraphic \newcounter\connectedwordcounter \starttexdefinition unexpanded fromword [#1]#2 \expandafter\pushmacro\csname cwc>#1\endcsname % nesting hack \doglobal\increment\connectedwordcounter \setxvalue{cwc>#1}{\connectedwordcounter}% \startpositionoverlay{highlightoverlay} \setMPpositiongraphic {\getvalue{cwc>#1}>f} {highlightgraphic} {from=\getvalue{cwc>#1}>f,to=\getvalue{cwc>#1}>t} \stoppositionoverlay \hpos{\getvalue{cwc>#1}>f}{#2}% \stoptexdefinition \starttexdefinition unexpanded toword [#1]#2 \hpos{\getvalue{cwc>#1}>t}{#2} \expandafter\popmacro\csname cwc>#1\endcsname % nesting hack \stoptexdefinition \setuppapersize[A6] \setupbackgrounds[page][background=highlight] \starttext \fromword[foo]{this} belongs to \toword[foo]{that} \fromword[bar]{this} belongs to \toword[bar]{that} \fromword[foo]{this} belongs to \toword[foo]{that} \fromword[bar]{this} belongs \fromword[bar]{this} belongs to \toword[bar]{that} to \toword[bar]{that} \stoptext ----------------------------------------------------------------- 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 -----------------------------------------------------------------

Aditya Mahajan

8:28 p.m.

On Tue, 3 Mar 2015, Hans Hagen wrote:

...

On 3/3/2015 6:12 PM, Aditya Mahajan wrote:

...
2. The metafun alternative is a bit inconvenient. Suppose you want to draw a bunch of such graphics (words connected by arrows). Then, you will need to ensure that node names are unique. (Is there a way to get around that?)

\newcounter\connectedwordcounter

\starttexdefinition unexpanded fromword [#1]#2 \expandafter\pushmacro\csname cwc>#1\endcsname % nesting hack \doglobal\increment\connectedwordcounter \setxvalue{cwc>#1}{\connectedwordcounter}% \startpositionoverlay{highlightoverlay} \setMPpositiongraphic {\getvalue{cwc>#1}>f} {highlightgraphic} {from=\getvalue{cwc>#1}>f,to=\getvalue{cwc>#1}>t} \stoppositionoverlay \hpos{\getvalue{cwc>#1}>f}{#2}% \stoptexdefinition

\starttexdefinition unexpanded toword [#1]#2 \hpos{\getvalue{cwc>#1}>t}{#2} \expandafter\popmacro\csname cwc>#1\endcsname % nesting hack \stoptexdefinition

Thanks. Aditya

Jörg Weger

7 Mar 7 Mar

1:08 p.m.

Thank you very much Hans and Aditya for your efforts and sorry for only replying now — momentarily I am writing a paper (of course typeset with ConTeXt :-) ) that has to get finished. For now I have dropped the idea of using that certain type of illustration in the paper as your METAPOST examples are working, but I had forgotten to say that I need effectively two arrows from one part of a sentence two two others (I am trying to illustrate a linguistic ambiguity problem) and I did not manage to get that working starting from your examples. Then I need a shape other than a tight frame to surround the words, TikZ’ rounded rectangles were too nice for that. In the moment I don’t have the time to dive deeper into METAPOST which seems to be a mighty beast from what I see in the Metafun manual, but in future I shurely will. The nice thing about TikZ are that you can use the self explaining commands rather intuitively (at least having some experience in Lilypond and now ConTeXt), that there are many different building blocks predefined that you can use and that there are hundreds of examples in the internet that you can use as starting point. So I managed to get some things working after a self-tought one day crash course – Metapost’s learnig curve seems to be a lot steeper to me. It is really a pity that ConTeXt mkiv does not save the node coordinates correctly as Wolfgang found out. Greetings Jörg On 03.03.2015 20:28, Aditya Mahajan wrote:

...

On Tue, 3 Mar 2015, Hans Hagen wrote:

...
On 3/3/2015 6:12 PM, Aditya Mahajan wrote:

...
2. The metafun alternative is a bit inconvenient. Suppose you want to draw a bunch of such graphics (words connected by arrows). Then, you will need to ensure that node names are unique. (Is there a way to get around that?)

\newcounter\connectedwordcounter

\starttexdefinition unexpanded fromword [#1]#2 \expandafter\pushmacro\csname cwc>#1\endcsname % nesting hack \doglobal\increment\connectedwordcounter \setxvalue{cwc>#1}{\connectedwordcounter}% \startpositionoverlay{highlightoverlay} \setMPpositiongraphic {\getvalue{cwc>#1}>f} {highlightgraphic} {from=\getvalue{cwc>#1}>f,to=\getvalue{cwc>#1}>t} \stoppositionoverlay \hpos{\getvalue{cwc>#1}>f}{#2}% \stoptexdefinition

\starttexdefinition unexpanded toword [#1]#2 \hpos{\getvalue{cwc>#1}>t}{#2} \expandafter\popmacro\csname cwc>#1\endcsname % nesting hack \stoptexdefinition

Thanks.

Aditya ___________________________________________________________________________________

If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : http://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Hans Hagen

3:52 p.m.

On 3/7/2015 1:08 PM, Jörg Weger wrote:

...

Metapost’s learnig curve seems to be a lot steeper to me. It is really a pity that ConTeXt mkiv does not save the node coordinates correctly as Wolfgang found out.

Is that really an mkiv issue? If I understood well tikz has its own way of keeping track of positions independent of context. 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 -----------------------------------------------------------------

Jörg Weger

6:30 p.m.

...

Is that really an mkiv issue? If I understood well tikz has its own way of keeping track of positions independent of context.

Normally yes. But there is TikZ’ “remember picture” mode. You can produce several independent TikZ pictures on a ConTeXt/LaTeX/plain text page. TikZ tells ConTeXt (or LaTeX or plain TeX) to give back the coordinates of those pictures on the page after the page has been typeset. TikZ can then draw interconnections between those images which are rendered in an extra run. In my example the rounded rectangles with the words inside are TikZ images which are put inline into the ConteXt text by the command \tikz[baseline] \node[<form of node>, draw, anchor=text] {<text content>} When I use \tikz[baseline, remember picture] … instead I activate the mode explained above and the arrows can be drawn according to the “remembered” coordinates after ConTeXt’s first typesetting run. According to Wolfgang Schuster in the first reply in this thread when using ConTeXt mkiv – other than mkii – file the coordinates given back after the first typesetting run look strange in TikZ’ auxiliary so there seems to be a mistake somnewhere between ConTeXt and TikZ in this regard. Greetings Jörg On 07.03.2015 15:52, Hans Hagen wrote:

...

On 3/7/2015 1:08 PM, Jörg Weger wrote:

...
Metapost’s learnig curve seems to be a lot steeper to me. It is really a pity that ConTeXt mkiv does not save the node coordinates correctly as Wolfgang found out.

Is that really an mkiv issue? If I understood well tikz has its own way of keeping track of positions independent of context.

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 ----------------------------------------------------------------- ___________________________________________________________________________________

If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : http://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Wolfgang Schuster

7:22 p.m.

...

Am 07.03.2015 um 18:30 schrieb Jörg Weger :

...
Is that really an mkiv issue? If I understood well tikz has its own way of keeping track of positions independent of context.

Normally yes.

But there is TikZ’ “remember picture” mode. You can produce several independent TikZ pictures on a ConTeXt/LaTeX/plain text page. TikZ tells ConTeXt (or LaTeX or plain TeX) to give back the coordinates of those pictures on the page after the page has been typeset. TikZ can then draw interconnections between those images which are rendered in an extra run.

Tikz uses the pdftex commands \pdfsavepos, \pdflastxpos and \pdflastypos to save the position of the graphics on the page. ConTeXt isn’t involved in this process which provides its own interface for this mechanism but Tikz doesn’t use it.

...

In my example the rounded rectangles with the words inside are TikZ images which are put inline into the ConteXt text by the command

\tikz[baseline] \node[<form of node>, draw, anchor=text] {<text content>}

When I use

\tikz[baseline, remember picture] …

instead I activate the mode explained above and the arrows can be drawn according to the “remembered” coordinates after ConTeXt’s first typesetting run.

According to Wolfgang Schuster in the first reply in this thread when using ConTeXt mkiv – other than mkii – file the coordinates given back after the first typesetting run look strange in TikZ’ auxiliary so there seems to be a mistake somnewhere between ConTeXt and TikZ in this regard.

The problem is the \write command which differs in MkII and MkIV and Tikz way to prevent the expansion of a command which is written to the auxiliary file. The following example demonstrates the difference between MkII and MkIV. In MkII the \noexpand prevents the expansion of \dummytext when writes content is written to the external file but in MkIV \write is redefined and this doesn’t work anymore. %%%% begin example \newwrite\testwrite \starttext \immediate\openout\testwrite=\jobname-dummytext.tmp \def\dummytext{First sample text.} \immediate\write\testwrite{\noexpand\dummytext} \def\dummytext{Second sample text.} \immediate\write\testwrite{\noexpand\dummytext} \immediate\closeout\testwrite \stoptext %%%% end example Wolfgang

Hans Hagen

8 Mar 8 Mar

11:11 a.m.

On 3/7/2015 7:22 PM, Wolfgang Schuster wrote:

...

...
Am 07.03.2015 um 18:30 schrieb Jörg Weger :

...
Is that really an mkiv issue? If I understood well tikz has its own way of keeping track of positions independent of context.

Normally yes.

But there is TikZ’ “remember picture” mode. You can produce several independent TikZ pictures on a ConTeXt/LaTeX/plain text page. TikZ tells ConTeXt (or LaTeX or plain TeX) to give back the coordinates of those pictures on the page after the page has been typeset. TikZ can then draw interconnections between those images which are rendered in an extra run.

Tikz uses the pdftex commands \pdfsavepos, \pdflastxpos and \pdflastypos to save the position of the graphics on the page. ConTeXt isn’t involved in this process which provides its own interface for this mechanism but Tikz doesn’t use it.

...
In my example the rounded rectangles with the words inside are TikZ images which are put inline into the ConteXt text by the command

\tikz[baseline] \node[<form of node>, draw, anchor=text] {<text content>}

When I use

\tikz[baseline, remember picture] …

instead I activate the mode explained above and the arrows can be drawn according to the “remembered” coordinates after ConTeXt’s first typesetting run.

According to Wolfgang Schuster in the first reply in this thread when using ConTeXt mkiv – other than mkii – file the coordinates given back after the first typesetting run look strange in TikZ’ auxiliary so there seems to be a mistake somnewhere between ConTeXt and TikZ in this regard.

The problem is the \write command which differs in MkII and MkIV and Tikz way to prevent the expansion of a command which is written to the auxiliary file.

The following example demonstrates the difference between MkII and MkIV.

In MkII the \noexpand prevents the expansion of \dummytext when writes content is written to the external file but in MkIV \write is redefined and this doesn’t work anymore.

%%%% begin example \newwrite\testwrite

\starttext

\immediate\openout\testwrite=\jobname-dummytext.tmp

\def\dummytext{First sample text.}

\immediate\write\testwrite{\noexpand\dummytext}

\def\dummytext{Second sample text.}

\immediate\write\testwrite{\noexpand\dummytext}

\immediate\closeout\testwrite

\stoptext %%%% end example

this should help: \unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi} (in future version i might replace write completely) 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 -----------------------------------------------------------------

Wolfgang Schuster

12:26 p.m.

...

Am 08.03.2015 um 11:11 schrieb Hans Hagen :

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong. %%%% begin example \newwrite\testwrite \immediate\openout\testwrite=\jobname-testwrite.tmp \parindent=1cm \starttext \indent\pdfsavepos\write\testwrite{1:\number\pdflastxpos:\number\pdflastypos} \noindent\pdfsavepos\write\testwrite{2:\number\pdflastxpos:\number\pdflastypos} \page \indent\pdfsavepos\write\testwrite{3:\number\pdflastxpos:\number\pdflastypos} \noindent\pdfsavepos\write\testwrite{4:\number\pdflastxpos:\number\pdflastypos} \page \indent\pdfsavepos\write\testwrite{5:\number\pdflastxpos:\number\pdflastypos} \noindent\pdfsavepos\write\testwrite{6:\number\pdflastxpos:\number\pdflastypos} \stoptext %%%% end example Wolfgang

Hans Hagen

12:55 p.m.

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...

...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

...

%%%% begin example \newwrite\testwrite

\immediate\openout\testwrite=\jobname-testwrite.tmp

\parindent=1cm

\starttext

\indent\pdfsavepos\write\testwrite{1:\number\pdflastxpos:\number\pdflastypos}

\noindent\pdfsavepos\write\testwrite{2:\number\pdflastxpos:\number\pdflastypos}

\page

\indent\pdfsavepos\write\testwrite{3:\number\pdflastxpos:\number\pdflastypos}

\noindent\pdfsavepos\write\testwrite{4:\number\pdflastxpos:\number\pdflastypos}

\page

\indent\pdfsavepos\write\testwrite{5:\number\pdflastxpos:\number\pdflastypos}

\noindent\pdfsavepos\write\testwrite{6:\number\pdflastxpos:\number\pdflastypos}

\stoptext %%%% end example

Wolfgang

___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : http://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Wolfgang Schuster

1:04 p.m.

...

Am 08.03.2015 um 12:55 schrieb Hans Hagen :

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl mailto:pragma@wxs.nl>>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same): 1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023 Content of the external file from MkII (different values for the first and second position on each page): 1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000 Wolfgang

Hans Hagen

6:39 p.m.

On 3/8/2015 1:04 PM, Wolfgang Schuster wrote:

...

...
Am 08.03.2015 um 12:55 schrieb Hans Hagen mailto:pragma@wxs.nl>:

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl mailto:pragma@wxs.nl>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same):

1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023

Content of the external file from MkII (different values for the first and second position on each page):

1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000

ypositions can differ a bit because we use different fonts and so does the interlinespace 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 -----------------------------------------------------------------

Wolfgang Schuster

7:07 p.m.

...

Am 08.03.2015 um 18:39 schrieb Hans Hagen :

On 3/8/2015 1:04 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 12:55 schrieb Hans Hagen mailto:pragma@wxs.nl>:

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl mailto:pragma@wxs.nl>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same):

1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023

Content of the external file from MkII (different values for the first and second position on each page):

1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000

ypositions can differ a bit because we use different fonts and so does the interlinespace

But the values for both positions should be different which isn’t the case for MkIV and the first page saves 0 for the x and y positions. Wolfgang

Hans Hagen

8:04 p.m.

On 3/8/2015 7:07 PM, Wolfgang Schuster wrote:

...

...
Am 08.03.2015 um 18:39 schrieb Hans Hagen :

On 3/8/2015 1:04 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 12:55 schrieb Hans Hagen mailto:pragma@wxs.nl>:

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl mailto:pragma@wxs.nl>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same):

1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023

Content of the external file from MkII (different values for the first and second position on each page):

1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000

ypositions can differ a bit because we use different fonts and so does the interlinespace

But the values for both positions should be different which isn’t the case for MkIV and the first page saves 0 for the x and y positions.

hm, after some playing with it i think that is a luatex bug (it expands the positions in a tex.write which is wrong so we need to revert to this: \def\syst_write_execute#1% {\ctxcommand{execute(\!!bs#1\!!es)}} \unexpanded\def\write#1#% {\ifnum#1=18 \expandafter\syst_write_execute \else \normalwrite#1% \fi} (context itself doesn't suffer from this as it does things differently; maybe i should make an api and then nil \pdf*pos) 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 -----------------------------------------------------------------

Hans Hagen

11:17 p.m.

On 3/8/2015 7:07 PM, Wolfgang Schuster wrote:

...

...
Am 08.03.2015 um 18:39 schrieb Hans Hagen :

On 3/8/2015 1:04 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 12:55 schrieb Hans Hagen mailto:pragma@wxs.nl>:

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl mailto:pragma@wxs.nl>:

this should help:

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else % the \detokenize makes sure we don't expand \noexpanded macros \normalwrite#1{\detokenize{#2}}% \fi}

(in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same):

1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023

Content of the external file from MkII (different values for the first and second position on each page):

1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000

ypositions can differ a bit because we use different fonts and so does the interlinespace

But the values for both positions should be different which isn’t the case for MkIV and the first page saves 0 for the x and y positions.

It's a side effect of expansion at the wrong time (where such primitives can return wrong values but at least they return something) but something like this works: \def\syst_write#1#2% {\ctxcommand{write(\number#1,\!!bs\normalunexpanded{#2}\!!es)}} \unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else \normalwrite#1{#2}% \fi} -- ----------------------------------------------------------------- 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 -----------------------------------------------------------------

Jörg Weger

13 Mar 13 Mar

2:17 p.m.

After putting the definition you proposed last in front of my minimal example in mkiv beta 2015.02.03 it worked once and then not any more. I also tried to remove all auxiliary files after a new pass. But today I installed beta 2015.03.10 (complete fresh install as I keep the older version in a renamed folder) and now the MWE with your extra code renders repeatedly correct. So this seems to be a workaround until you rewrite “write” as you said. I will experiment further with tikz’ ‘remember’ picture mode. Greetings Jörg %%%%%%%%%%%%%%%% START MWE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %D %%%%%%% Hans’ redefinition of write function: %%%%%%%%%%% \def\syst_write#1#2% {\ctxcommand{write(\number#1,\!!bs\normalunexpanded{#2}\!!es)}} \unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else \normalwrite#1{#2}% \fi} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \usemodule[tikz] \usetikzlibrary[shapes.misc,arrows] \starttext \tikz[baseline,remember picture] \node[rounded rectangle,draw,anchor=text] (n1) {this}; belongs to \tikz[baseline,remember picture] \node[rounded rectangle,draw,anchor=text] (n2) {that}; \starttikzpicture [remember picture,overlay] \draw[->] (n1.north) to [bend left] (n2.north); \stoptikzpicture \stoptext %%%%%%%%%%%%%%%% STOP MWE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% On 08.03.2015 23:17, Hans Hagen wrote:

...

On 3/8/2015 7:07 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 18:39 schrieb Hans Hagen :

On 3/8/2015 1:04 PM, Wolfgang Schuster wrote:

...
...
Am 08.03.2015 um 12:55 schrieb Hans Hagen mailto:pragma@wxs.nl>:

On 3/8/2015 12:26 PM, Wolfgang Schuster wrote:

...
> Am 08.03.2015 um 11:11 schrieb Hans Hagen mailto:pragma@wxs.nl > mailto:pragma@wxs.nl>: > > this should help: > > \unexpanded\def\writeviatex#1#2% > {\ifx\normalwrite\relax\else > % the \detokenize makes sure we don't expand \noexpanded macros > \normalwrite#1{\detokenize{#2}}% > \fi} > > (in future version i might replace write completely)

Only partially because the saved positions from \pdfsavepos are wrong.

in what sense wrong?

Content of the external file from MkIV (the entries on the first page show 0 and the values on page 2 and 3 are always the same):

1:0:0 2:0:0 3:4661756:45255023 4:4661756:45255023 5:4661756:45255023 6:4661756:45255023

Content of the external file from MkII (different values for the first and second position on each page):

1:6526435:46204089 2:4661756:45256000 3:6526435:46204089 4:4661756:45256000 5:6526435:46204089 6:4661756:45256000

ypositions can differ a bit because we use different fonts and so does the interlinespace

But the values for both positions should be different which isn’t the case for MkIV and the first page saves 0 for the x and y positions.

It's a side effect of expansion at the wrong time (where such primitives can return wrong values but at least they return something) but something like this works:

\def\syst_write#1#2% {\ctxcommand{write(\number#1,\!!bs\normalunexpanded{#2}\!!es)}}

\unexpanded\def\writeviatex#1#2% {\ifx\normalwrite\relax\else \normalwrite#1{#2}% \fi}

Wolfgang Schuster

2:32 p.m.

...

Am 13.03.2015 um 14:17 schrieb Jörg Weger :

After putting the definition you proposed last in front of my minimal example in mkiv beta 2015.02.03 it worked once and then not any more. I also tried to remove all auxiliary files after a new pass.

But today I installed beta 2015.03.10 (complete fresh install as I keep the older version in a renamed folder) and now the MWE with your extra code renders repeatedly correct.

So this seems to be a workaround until you rewrite “write” as you said.

I will experiment further with tikz’ ‘remember’ picture mode.

1. The current beta has a fix for the \write command and you don’t need Hans patch. 2. Commands with a _ in the name, e.g. \syst_write can only be used when you place them in a group which starts with \unprotect and ends with \protect. Wolfgang

3674

Age (days ago)

3684

Last active (days ago)

List overview

Download

24 comments

7 participants

participants (7)

Aditya Mahajan
Hans Hagen
Jörg Weger
Jörg Weger
Mojca Miklavec
Ulrike Fischer
Wolfgang Schuster