Sanjoy Mahajan wrote:

...

...

It would be nice if placefigure can be able to detect if the best place is the top or the bottom of the page accordingly to its call,

A related point, which I didn't fiugre out how for my own TeX figure-placement macros, is correct sequential numbering when some figures are at the top of the page. Some pseudocode on one page:

=============== snip =================== \ref{figure:narrow} is a narrow figure.

\figuremacro{figure:narrow}{narrowfig.1}{A narrow figure.}

\ref{figure:wide} is a wide figure.

\figuremacro{figure:wide}{widefig.1}{A wide figure} =============== snip ===================

Here \figuremacro figures out (sorry) that narrow figures go in the margin and wide figures go at the top of the page. But the page will look funny: The narrow figure will be numbered, say, Figure 10, and the wide figure will be Figure 11, but Figure 11 will appear before Figure 10 on the page. Which is disconcerting to the reader.

One solution is to look at all the figures on the page and then number them, instead of numbering them when \figuremacro is executed. I couldn't figure out how to do that in my TeX macros, but maybe there are hooks into the context output routine for such tricks?

maybe \setupfloats[numbering=nocheck] does what you want -- ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Sanjoy Mahajan wrote:

Hans Hagen wrote:

...

maybe \setupfloats[numbering=nocheck] does what you want

I think I want the opposite. The documentation for \setupfloats at http://texshow.contextgarden.net/ says

New option: numbering yes nocheck (default yes) nocheck causes no renumbering so it preserves figure's source order

My problem was that the source order (which is the order of first reference, probably) is not the order on the page. The wide figure floats to the top of the page (because it's wide) even though its source position is later than the narrower figure that fits in the margin.

So 'numbering=yes' will number according to position on the page?

no, actually floats will be renumbered automatically unless nocheck is given; the problem is that the order in which margins etc flush plays a role then. I can imagine another solution but i need a dark rainy non-busy day for that Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Taco Hoekwater wrote:

Hans Hagen wrote:

...

...

So 'numbering=yes' will number according to position on the page?

no

To avoid confusion:

When 'page' refers to the typeset result, then Hans' answer actually means 'yes'. Or it should, at least. I remember files where I had to run texexec twice though. I am not sure if that bug still happens, so be careful.

Right, I meant in pages of the typeset result. I'll watch for the need to run texexec and make a minimal example if I find the problem. Meanwhile I've been experimenting in baby steps, and numbering=yes does what I want, as does specifying nothing about numbering. But numbering=nocheck doesn't stop it, so I'm confused by what the nocheck option does (doesn't cause me problems since the default is what I want, but I'm curious). Probaby I'm just using an old version of context since the texshow-web says that numbering=nocheck is a new option. My last attempt at upgrading didn't go well (I have tetex-3.0) but maybe it's time to try again. $ texexec --version TeXExec 5.2.4 - ConTeXt / PRAGMA ADE 1997-2005 texexec : TeXExec 5.2.4 - ConTeXt / PRAGMA ADE 1997-2005 texutil : TeXUtil 9.0.0 - ConTeXt / PRAGMA ADE 1992-2004 tex : pdfeTeX, 3.141592-1.21a-2.2 (Web2C 7.5.4) context : ver: 2005.01.31 cont-en : ver: 2005.01.31 fmt: 2006.3.28 mes: english With this output on the file $ texexec 2m [...snip...] systems : begin file 2m at line 6 (/usr/share/texmf-tetex/tex/context/sample/tufte.tex) figures : dimensions of ./a.1 loaded from figurefile itself (./2m.tuo) floatblocks : 1 renumbered / figure 1 => 2 floatblocks : 1 placed (/usr/share/texmf-tetex/tex/context/sample/knuth.tex) figures : dimensions of ./a.2 loaded from figurefile itself floatblocks : 2 renumbered / figure 2 => 1 floatblocks : 2 placed [1.1] systems : end file 2m at line 13 [...snip...] So the figures get renumbered. Here's the 2m.tex file: ====================================================================== \setuplayout[textwidth=3.6in, rightmargin=2in] \useexternalfigure[mpg][a.1] \useexternalfigure[mpg2][a.2] \setupfloats[numbering=nocheck] \starttext \input tufte \placefigure[rightmargin]{first in source file.}{\externalfigure[mpg]} \par \input knuth \placefigure[top]{second in source file.}{\externalfigure[mpg2]} \par \stoptext ====================================================================== Where a.mp to produce a.1 and a.2 is: ====================================================================== beginfig(1) fill unitsquare scaled 1in withcolor red; endfig; beginfig(2) fill unitsquare scaled 1in withcolor green; endfig; end ====================================================================== -Sanjoy `A society of sheep must in time beget a government of wolves.' - Bertrand de Jouvenal

From: Hans Hagen

\useMPlibrary[dum] \setupfloats[numbering=nocheck] \starttext \input tufte \placefigure[rightmargin]{first} {\externalfigure[dummy]} \par \input knuth \placefigure[top] {second}{\externalfigure[dummy]} \par \stoptext

does commenting the no check line makes a differnece?

nocheck: top=2, margin=1 yes : top=1, margin=2 (default behaviour)

No difference. I get the default behavior in both cases, and both log files mention the renumbering: $ texexec 3.tex > somewhere $ mv 3.log 3-nocheck.log [comment out the nocheck line] $ texexec 3.tex > somewhereelse $ grep renumber 3{,-nocheck}.log 3.log:floatblocks : 1 renumbered / figure 1 => 2 3.log:floatblocks : 2 renumbered / figure 2 => 1 3-nocheck.log:floatblocks : 1 renumbered / figure 1 => 2 3-nocheck.log:floatblocks : 2 renumbered / figure 2 => 1 TeXExec 5.2.4 - ConTeXt / PRAGMA ADE 1997-2005 texexec : TeXExec 5.2.4 - ConTeXt / PRAGMA ADE 1997-2005 texutil : TeXUtil 9.0.0 - ConTeXt / PRAGMA ADE 1992-2004 tex : pdfeTeX, 3.141592-1.21a-2.2 (Web2C 7.5.4) context : ver: 2005.01.31 cont-en : ver: 2005.01.31 fmt: 2006.3.28 mes: english Maybe this context installation is too old for nocheck. On the other hand, it should complain then? -Sanjoy

Hans Hagen wrote:

Sanjoy Mahajan wrote:

...

context : ver: 2005.01.31 cont-en : ver: 2005.01.31 fmt: 2006.3.28 mes: english

Maybe this context installation is too old for nocheck. On the other hand, it should complain then?

i remember a bug (actually the mechanism is rather old, so the bug must have been a temp one)

I tried it on context live (2006.04.04), and it works as hoped (figure 2 is the on on top of the page with numbering=nocheck, otherwise it's numbered in page order). Definitely I need to do an upgrade. But I noticed another problem: The context-live pdf is black/white, whereas on my system the figures come out in color. I could reproduce the difference with the following: $ texexec --pdf 3 => 3.pdf is B/W versus $ texexec 3.tex $ dvips 3 $ ps2pdf 3.ps => 3.pdf is in color texexec --color makes the two pathways give identical pdf files, but shouldn't they also be identical (and B/W) without the --color? Or is that too much of a mess, to inquire into what colors the metapost (or other EPS) figures are doing. -Sanjoy `A society of sheep must in time beget a government of wolves.' - Bertrand de Jouvenal

Sanjoy Mahajan wrote:

...

...

It would be nice if placefigure can be able to detect if the best place is the top or the bottom of the page accordingly to its call,

A related point, which I didn't fiugre out how for my own TeX figure-placement macros, is correct sequential numbering when some figures are at the top of the page. Some pseudocode on one page:

=============== snip =================== \ref{figure:narrow} is a narrow figure.

\figuremacro{figure:narrow}{narrowfig.1}{A narrow figure.}

\ref{figure:wide} is a wide figure.

\figuremacro{figure:wide}{widefig.1}{A wide figure} =============== snip ===================

Here \figuremacro figures out (sorry) that narrow figures go in the margin and wide figures go at the top of the page. But the page will look funny: The narrow figure will be numbered, say, Figure 10, and the wide figure will be Figure 11, but Figure 11 will appear before Figure 10 on the page. Which is disconcerting to the reader.

this 'choose between locations' feature is kind of present but only with 'here' as fallback but it's trivial to make that configurable: % first a new key \startconstants dutch english german czech italian romanian french fallback: terugval fallback fallback fallback fallback fallback fallback \stopconstants % use that key \long\def\dofloat#1#2#3#4% {\dosetfloatbox{#1}{#2}{#3}{#4}% \doifelsevaluenothing{\??fl#4\c!criterium} {\dogetfloatbox{#1}\empty} {\ifdim\wd\floatbox>\getvalue{\??fl#4\c!criterium}\relax \postcenterfloatbox{\wd\floatbox}% else we get left aligned %dogetfloatbox{#1}\v!here % see details/pascal \dogetfloatbox{#1}{\executeifdefined{\??fl#4\c!fallback}\v!here}% PATCHED \else \dogetfloatbox{#1}\empty \fi}} % pass the new location (will be done differently, hack \def\dogetfloatbox#1#2% {\ifvisible \doifelsenothing{#2} {\getfromcommalist[#1][1]% \@EA\beforesplitstring\commalistelement\at:\to\floatmethod \@EA\aftersplitstring \commalistelement\at:\to\floatcolumn \@EA\aftersplitstring \floatcolumn\at*\to\floatrow \@EA\beforesplitstring\floatcolumn\at*\to\floatcolumn % todo: nog algemeen otr \ifx\OTRSETsetpreferedcolumnslot\undefined\else \OTRSETsetpreferedcolumnslot\floatcolumn\floatrow \fi} {\let\floatcolumn\empty \let\floatrow\empty \edef\floatmethod{#2}}% \doifundefined{\string\floatmethod\floatmethod} {\let\floatmethod\v!here}% \getvalue{\string\floatmethod\floatmethod}[\floatmethod,#1]% PATCHED \fi} \setupfloats[numbering=nocheck] % we use the marginwidth as criterium \setupfloat[figure][criterium=\marginwidth,fallback=bottom] \starttext \placefigure[bottom]{}{} \input tufte \placefigure[left]{}{} \input tufte \placefigure[left]{}{\framed[width=.8\marginwidth]{}} \input tufte \stoptext (we needed that feature a couple of years ago in a project -) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Sanjoy Mahajan wrote:

Taco's msg about no meaning yes made me wonder: what's the special case or problem you were thinking of?

probably that the automatic renumbering sometimes interferes with non trivial sequential flushing

Perhaps Holland in the winter can offer the dark, rainy day (just like here in Boston, USA) but the 'non-busy' may be more difficult.

i remember a pretty hot boston made even more hot because i had to carry the nice big book "inventing the charles river" around -) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------