[tex-k] a bug in dvipsk

Stepan Kasal kasal at ucw.cz
Tue May 18 12:59:06 CEST 2004

  a bug in dvips was reported on a Czech mailing list.
The recent info and man pages seem to introduce option -y, a mag.
factor for y-axis.

When I look into the code, I cannot imagine that it could work
under any circumstances:

The option reading code says (shortened):

case 'x' : case 'y' :
               *p || p = argv[++i] ;
               sscanf(p, "%lg", &mag);
               overridemag = (c == 'x' ? 1 : -1) ;

The configuration is then used this way (prescan.c);

   if (overridemag > 0) (void)signedquad() /* throw away */;
   else if (overridemag < 0) mag = (mag * signedquad() + 500.0) / 1000.0 ;
   else mag = signedquad() ;

I don't understand the meaning of the middle expression.

And, of course, there in no provision for handling combination of
both -x and -y.

What about squashing the -y altogether, both from docs and source?

Have a nice day,
	Stepan Kasal

More information about the tex-k mailing list