Return-Path: Received: from [62.76.169.38] (HELO video) by vsu.ru (CommuniGate Pro SMTP 3.5b5) with ESMTP id 3935348; Fri, 12 Oct 2001 17:27:09 +0400 To: "Berthold K.P. Horn" Cc: CyrTeX-en@vsu.ru, CyrTeX-ru@vsu.ru, tex-fonts@math.utah.edu, tetex@informatik.uni-hannover.de Subject: Re: CM-Super font package v0.3.0 released References: <5.1.0.14.2.20011012082044.02f9ede0@mail.ai.mit.edu> From: Vladimir Volovich Date: Fri, 12 Oct 2001 17:27:29 +0400 In-Reply-To: <5.1.0.14.2.20011012082044.02f9ede0@mail.ai.mit.edu> ("Berthold K.P. Horn"'s message of "Fri, 12 Oct 2001 08:25:47 -0400") Message-ID: Lines: 29 User-Agent: Gnus/5.090004 (Oort Gnus v0.04) Emacs/21.0.103 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii "BKPH" == Berthold K P Horn writes: >> * Generated precise (non-integer) widths for all glyphs in PFB >> files to match TFM widths (using continuous fraction >> approximation). Corrected width of space glyph to match default >> TeX font space width. BKPH> It's nice to see this rather subtle and unappreciated bit of BKPH> cleverness from the original BSR/Y&Y CM fonts replicated :-) thank you. :-) BKPH> Using continued fractions you can get about n + m bits worth of BKPH> accuracy when the numerator and denominator have n and m bits BKPH> respectively. Much better then the obvious way of dividing by BKPH> some fixed quantity like 1000, where you only get about n bits BKPH> worth... And of course much better than the junk fonts that BKPH> just round to the nearest integer. i've added the following paragraph into the README file: It should also be noted that the fonts use precise (non-integer) glyph widths which better match the TFM widths. These widths are generated using the best approximation (based on continuous fractions) with the denominator not exceeding 107 to fit in 1 byte in CharString. Apparently, such subtle technique was used first in BSR/Y&Y CM fonts. Best, v.