Quoting WN :

Hi Aditya

I tried your code and performed more tests. My first tests show your fix works. But than I added a \section{} in my test code and I am sorry to say that the math matrix alignment does not work at that point. However the extra math fonts I defined still do work.

Ok. I will test this to see what is happening. (I do not have an uptodate context on this machine).

General observation I thought from the examples like using the RSFS font on the wiki page http://wiki.contextgarden.net/rsfs that using other fonts for math should be easy. Thinking about this specific example I now understand that this example is flawed as some default math characters are also overwritten.

I am out of depths here. I do not really understand how math fonts are handled. The wiki example was modified from a suggestion from Taco. For rsfs, you can go to the plain tex way of doing things also. Aditya

PS Attached my modified test code along with the map file for the RSFS fonts. For the RSFS fonts I am using \definedfont and not using the definebodyfont.

...

Hi Wim,

On Tue, 31 Jul 2007, WN wrote:

see previous email

...

Phew! This was a hard to find bug. First, to illustrate the bug: \definetypeface[mainface] [rm] [serif] [modern] [default] [encoding=texansi] \definetypeface[mainface] [mm] [math] [modern] [modern] [encoding=texansi]

\setupbodyfont[mainface,11pt]

\loggingall

\starttext \startformula A = \startmatrix[align={left,right}] \NC 11 \NC 2 \NR \NC 2 \NC -5 \NR \stopmatrix \stopformula

\stoptext

Try with and without the font declarations. With the font declarations, you lost matrix alignment. The reason is that \startmatrix uses \scratchcounter, which gets rewritten when using a font collection other than "default". One fix it to replace \scractcounter in matrix constructions by some other counter, for example \!!counta.

So, add this to your file (beware of false line breaks in email)

\unprotect

\def\dostartmathmatrix[#1][#2]% {\begingroup \edef\currentmathmatrix{#1}% \doifassignmentelse{#2}{\setupmathmatrix[#1][#2]}\donothing \null

\executeifdefined{\??mx:\mathmatrixparameter\c!location}{\getvalue{\??mx:\v!lohi}}% \mathmatrixleft \mathmatrixbox\bgroup \pushmacro\domatrixNC \let\endmath\relax \def\NC{\domatrixNC}% \def\MC{\domatrixNC\ifmmode\else$\def\endmath{$}\fi}% \global\let\domatrixNC\dodomatrixNC \def\NR{\endmath\global\let\domatrixNC\dodomatrixNC\crcr}% \normalbaselines \mathsurround\zeropoint \everycr\emptytoks \tabskip\zeropoint \eqaligncolumn\zerocount % could be \!!counta

\processcommacommand[\mathmatrixparameter\c!align]{\advance\eqaligncolumn\plusone\dosetmatrixcolumn}% \!!counta=\ifnum\eqaligncolumn>\!!counta \eqaligncolumn \else \plusone \fi \global\eqaligncolumn\plusone \preparemathmatrix } % uses !!counta

\def\buildmathmatrix % beware: etex only {\scratchtoks\emptytoks \expanded{\scratchtoks{\the\scratchtoks\the\!!toksa}}% \dorecurse{\numexpr\!!counta-\plusone\relax} {\expanded{\scratchtoks{\the\scratchtoks\the\!!toksb}}}% \expanded{\scratchtoks{\the\scratchtoks\the\!!toksc }}}

\protect

Hans, this needs to be changed in core-mat.tex. I do not know if something can be done to avoid such errors in future.

Wim, I am attaching a modified version of your file which gives the correct output. (I needed to make some changes since I did not have your map file for rsfs). You loose some symbols from the math collection. I do not know how to add extra math symbol sets without loosing these characters. Look at math-tim.tex to see how more families can be supported. But I would suggest waiting for mkiv, where the limit on font families will be removed and some of this font mess would be cleaner.

Aditya

Hi Aditya I tried the new code you send me, and of course I did a quick test. But when doing a texmfstart texexec --autopdf --pdf testalignfontnew.tex texexec hangs ''forever'' while trying to process the \section{} command. Actually its pdftex trying to do something. To make sure I did not make a edit mistake, I copied my old test code into the file testalignfontnew.tex and run a diff to see where the difference with the new code are. Looking at the code (and I am not a Tex expert, and forgive me, when I ask a dumb question) I don't see where the \scratchcounter itself is defined ? Must surely be a global Context counter ? Attached the new test code. Hope it helps the debugging process and thanks for all your help sofar. Wim

Hi Wim,

Quoting WN :

...

I tried your code and performed more tests. My first tests show your fix works. But than I added a \section{} in my test code and I am sorry to say that the math matrix alignment does not work at that point.

:-) Section modifies \!!counta so things get messed up again. Actually, this really helped to narrow down the bug. Basically, dostartmatrix does a test

\scratchcounter=\ifnum\eqaligncolumn>\scratchcounter \eqaligncolumn \else \plusone \fi

which is comparing eqaligncolmn with scratch counter, but scratchcounter is never initialized. Therefore, everything depends on what was the value of scratchcounter, which explains, different behaviour depending on when the matrix occurred, and why you had a hard time pinning down a minimal example.

The fix is easy, change the scratchcounter to zerocount (or maybe just change to \ifcase\eqaligncolunm). So add this to your file,

\unprotect \def\dostartmathmatrix[#1][#2]% {\begingroup \edef\currentmathmatrix{#1}% \doifassignmentelse{#2}{\setupmathmatrix[#1][#2]}\donothing \null

\executeifdefined{\??mx:\mathmatrixparameter\c!location}{\getvalue{\??mx:\v!lohi}}% \mathmatrixleft \mathmatrixbox\bgroup \pushmacro\domatrixNC \let\endmath\relax \def\NC{\domatrixNC}% \def\MC{\domatrixNC\ifmmode\else$\def\endmath{$}\fi}% \global\let\domatrixNC\dodomatrixNC \def\NR{\endmath\global\let\domatrixNC\dodomatrixNC\crcr}% \normalbaselines \mathsurround\zeropoint \everycr\emptytoks \tabskip\zeropoint \eqaligncolumn\zerocount \processcommacommand[\mathmatrixparameter\c!align]{\advance\eqaligncolumn\plusone\dosetmatrixcolumn}% \scratchcounter=\ifnum\eqaligncolumn>\zerocount \eqaligncolumn \else \plusone \fi \global\eqaligncolumn\plusone \preparemathmatrix } % uses scratchcounter \protect

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 : https://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Hi Aditya, You don't have to check the font stuff. Everything works (alignment and the fonts) :-) Thanks for your all help. Wim

Hi Wim,

Quoting WN :

...

I tried the new code you send me, and of course I did a quick test. But when doing a texmfstart texexec --autopdf --pdf testalignfontnew.tex texexec hangs ''forever'' while trying to process the \section{} command. Actually its pdftex trying to do something.

To make sure I did not make a edit mistake, I copied my old test code into the file testalignfontnew.tex and run a diff to see where the difference with the new code are.

Looking at the code (and I am not a Tex expert, and forgive me, when I ask a dumb question) I don't see where the \scratchcounter itself is defined ? Must surely be a global Context counter ?

Yes. \scratchcounter is defined in syst-gen (IIRC). So are a bunch of other temporary variables.

...

Attached the new test code. Hope it helps the debugging process and thanks for all your help sofar.

I can not really check the font stuff, since I do not have txfonts and rsfs on this machine. But the matrix alignment appears to be working fine. I am attaching the test for just the matrix alignment. When I get home, I will try to see what goes wrong when you add other fonts.

Aditya