Chris Rowley <[log in to unmask]> writes on Mon, 5 Feb 2001 23:26:16 +0000: >> ... >> ... since the fonts available are going to get more and more >> diverse (if only very slowly) the robust medium-term solution for a >> TeX-like typesetting engine is to add the ability to bundle the >> font-metrics with the document (possibly virtually when network >> font warehouses really exist). >> ... To some extent, TeX already takes care of this, in that it records a font checksum in the DVI file that the DVI driver should match against the fonts used, and report any discrepancies. Of course, when they do mismatch, the average user has no idea what to do about it. Adobe Portable Document Format (PDF) takes this approach of saving metrics for all fonts used. I have an example here on my desk of a PostScript output from earlier today using a font in which the round-trip from PostScript to PDF to PostScript resulted in a substitution of Adobe Sans MM in place of the original, Impact (a Microsoft TrueType poster font). This is both good and bad: the substitution happens silently, so the user is neither informed of it, or bothered by it. However, the two PostScript files definitely look different when printed; even their page dimensions changed. PDF was trying to ensure that if font substitutions occurred, the character spacing should be close to the original, even if the glyph appearance is changed. Some people find this objectionable: the U.S. National Science Foundation, for example, REQUIRES that PDF files for grant proposals contain embedded subsetted fonts to ensure uniform appearance everywhere. The notion of network font warehouses is, I suspect, far off, given the copyright and licensing issues on the vast majority of fonts. While Adobe has officially taken the position that their licensed fonts, if subsetted, can be legally embedded in distributed PDF and PostScript documents without further license fee payments, other vendors are not so accommodating: Bitstream's license makes even the font metrics in the .afm files confidential, e.g. >> Comment Copyright 1987-1992 as an unpublished work by Bitstream Inc., Cambridge, MA. >> Comment All rights reserved >> Comment Confidential and proprietary to Bitstream Inc. It is not clear what this means for PDF; a lawyer might conceivably argue that a Bitstream font cannot be legally used in a PDF file at all, because that constitutes copying of the font metrics. Whether this is a good business decision is another question :^). Thus, this brings us round to the alternative offered by PostScript to PDF conversion software, like Adobe Acrobat Distiller, and ghostscript ps2pdf, of forcing the embedding of all fonts used. The feature was undoubtedly added because customers wanted it, which suggests that if a descendant of TeX were to consider this issue, it ought to carefully examine the PDF experience. In closing, I note that the PDF-1.0 specification defined 14 base fonts (1993 edition of Portable Document Format Reference Manual, p. 64) which all PDF viewers are required to provide, so that they never need to be embedded. The PDF-1.3 specification (2000 second edition of same) no longer even has the notion of `base fonts'. Perhaps this is in recognition of a problem that has been discussed before on the tex-fonts list, that various flavors of the PDF Times serif base font from different vendors have not only had somewhat different glyph offerings, but also different metrics. Or perhaps it is because with 20K+ fonts on the market (see http://www.math.utah.edu/~beebe/fonts/fonts-to-vendors.html for a catalog), why should 14 of them be singled out for special treatment as `base fonts'. As more (La)TeX users learn how to expand their font choices beyond (Extended) Computer Modern, our community has to face the font availability problem too. ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: [log in to unmask] - - Department of Mathematics, 322 INSCC [log in to unmask] [log in to unmask] - - 155 S 1400 E RM 233 [log in to unmask] - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - -------------------------------------------------------------------------------