========================================================================= Date: Mon, 16 Dec 1991 13:52:04 +0100 Reply-To: The TUG DVI driver standards discussion list Sender: The TUG DVI driver standards discussion list From: "Eberhard Mattes" Subject: Zero interword space If a DVI processor uses TFM files, it has to use a function of the interword space parameter from the TFM file as word_space parameter (etc.) of the rounding algorithm. The interword space of cmmi10 (cmmi5, ...) is zero, therefore DVI coordinates are always rounded immediately to device coordinates without accumulating rounding errors until max_drift is reached. If TFM files are not used, a function of the design size is used as word_space parameter (etc.) of the rounding algorithm, therefore rounding errors are accumulated. As there are no `words' typeset from cmmi10 (italics correction is used between characters), there might be no difference between output created with and output created without TFM files. Has anyone thought about this? What's about accents? If max_drift should be > 0 for cmmi*, DVI processors should ignore the information from the TFM file if the interword space is zero. If max_drift should be 0 for cmmi*, DVI processors which don't use TFM files should use a list of fonts for which max_drift is to be set to zero. What's the correct behavior for zero interword space? Eberhard Mattes (mattes@azu.informatik.uni-stuttgart.de) ========================================================================= Date: Mon, 16 Dec 1991 15:08:20 MEZ Reply-To: The TUG DVI driver standards discussion list Sender: The TUG DVI driver standards discussion list From: Joachim Schrod Subject: Re: Zero interword space In-Reply-To: <9112161259.AA10989@HP5.ITI.INFORMATIK.TH-DARMSTADT.DE>; from "Eberhard Mattes" at Dec 16, 91 1:52 pm Eberhard Mattes wrote: > > If a DVI processor uses TFM files, it has to use a function of the > interword space parameter from the TFM file as word_space parameter (etc.) > of the rounding algorithm. A minor correction: It *may* use the word_space parameter (as outlined by you in the rest of your mail). > The interword space of cmmi10 (cmmi5, ...) is zero, therefore DVI > coordinates are always rounded immediately to device coordinates > without accumulating rounding errors until max_drift is reached. > > If TFM files are not used, a function of the design size is used as > word_space parameter (etc.) of the rounding algorithm, therefore > rounding errors are accumulated. > > [...] > > What's the correct behavior for zero interword space? Exactly as you outlined above. *If* parameters of the TFM files are used for the computation of word_space, *then* max_drift will be zero. Ie, there will be a difference in the output between drivers that use TFM files and drivers that do not use them. I do not think that we will be able to specify a reliable scheme that circumventes this. In addition, I'm against the idea of a `predefined font list.' Since these fonts will be from the CM family, this would open up this whole question if DVI is inherently tied to CM fonts. I don't think so. (I already see problems in the fact that bitmapped fonts are addressed so explicitely in the standard.) The explication of the max_drift method said This method for tracking the positions is oriented towards the typesetting of text. and continues with examples where it will not work properly. I might add another paragraph where I explain that it is not sufficient for math texts as well, and that it's usually better to use the TFM files. Any comments? -- Joachim // PLEASE NOTE CHANGE OF EMAIL ADDRESS !! =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Joachim Schrod Email: schrod@iti.informatik.th-darmstadt.de Secretary of TUG DVI Driver Standards Committee