[tex-k] a bug in dvipsk

Stepan Kasal kasal at ucw.cz
Wed May 19 14:26:33 CEST 2004


Hello Tom & Tom,
  actually this all leads to more problems.
There are several fields where an improved documentation could help
very much.

On Tue, May 18, 2004 at 12:08:10PM -0700, Tomas G. Rokicki wrote:
> -x means ignore dvi magnification and force magification to be
> the given value.  -y means multiply given dvi magnification by
> the given value.

Thanks for telling us.  But, as Cyril has pointed out, this is not
clearly described in the manpage/info.

  "-y num Set the y magnification ratio to num/1000.  See -x above."

When I hear "y magnification ratio" I inevitably think about y axe.

> It makes no sense to combine these two options; what would that mean?

One can imagine that  -x 1500 -y 2000  would yield a mag of 3000.
So you should perhaps document the current behaviour.

Ahnother problem is that -y means that the following line will be
executed:
        	mag = (mag * signedquad() + 500.0) / 1000.0 ;
instead of
	        mag = (mag * signedquad()) / 1000.0 ;

The variable `mag' is double now, so it's no longer necessary to
add that "+ 500.0" distortion.  Will you please fix it?

> Is it possible the original bug report was confused about -x/-y
> (which change dvi magnification, but always in an aspect
> preserving way) and -X/-Y (which specify the output device
> resolution when that resolution is not square)?

You are almost right.  Cyril has observed that laser printers often don't
have the exact declared resolution.  He wanted to correct this.

As his printer doesn't have PostScript, he solved it by specifying
a resolution to ghostscript, like -r1200x1208 or something.

If this problem were solved on the PostScript level, -X and -Y options to dvips
wouldn't probably be enough.  There are also .pfb fonts and the device _thinks_
that it has square pixels.  One would have to add a deformation on PostScript
level, but I'm afraid the pstools package doesn't have such feature.
If the dvips manual could contain some advieces on this, it would be very nice.

Thank you in advance for any help with these.
Have a nice day,
	Stepan



More information about the tex-k mailing list