9.4. Conclusions and Glimpses at the Future
In this chapter we have seen, without going too deeply into the details, various aspects of font management in TEX and Ω. Its apparent complexity is related to the clash of cultures that occurred between the initial philosophy of TEX (TFM metrics, on the one hand, and GF or PK bitmaps on the other) and the chaotic real world of the fonts of the past two decades. How can we get the most benefits from TEX font metrics—and from virtual fonts in particular? At the same time, how can we adapt fonts (PostScript, TrueType, OpenType, etc.) for use with TEX when they are a priori foreign to it?
We have seen that most of the problems stem from the incompatibility between these two philosophies and from the fact that we cannot provide more information to TEX than what exists within the concept of a virtual font. For instance, PostScript and TrueType fonts contain the notion of a bounding box, whereas virtual fonts know nothing about it. We are therefore obliged to perform all the operations involving a bounding box (e.g., precise placement of diacritical marks) while we still have access to this information—namely, at the time that the virtual font is created by fontinst, when it has just read the AFM files and accumulated the data for the bounding boxes. After that, it will be too late: TEX will never again have access to the information.
Another example, going in the opposite direction: ligatures. We have seen that when ligatures appear in PostScript ...