- /* Trying to fix this problem:
- *
- * I expand fontlab to the maximum of fonts troff can
- * handle. The maximum number i, due to the two chars
- * fontname limit, is 99.
- * If we don't use the (named) font in one of the
- * standard position, we install the name in the next
- * free slot. Whenever we need info about the font, we
- * read in the data at position zero, and secretly use
- * the data (actually only necessary for the width
- * and ligature info). The ptfont() routine will tell
- * the device filter to put the font always at position
- * zero, so no need to change these filters.
- * Yes, this is a bit kludgy.
- *
- * BTW, I removed the directory stuff completly since
- * setfp is always called as setfp(x, y, 0).
- *
- * TODO: It must be possible to do this with all the
- * font positions. Dynamically switch the ``mounted''
- * fonts in a LRU style, whithout bothering the user
- * with it. Then they never have to use the .fp
- * request. For compatibilty reason we could leave it
- * in (it can shuffle the date in the fontlab table),
- * or make it a NOOP.
- * Of course we do not touch positions with a ``Special'' font.
- *
- * Simple things TODO
- * A routine to extract two characters troff
- * names is handy.
- * A routine to find out whether a font discription
- * file is available, currently setfp(...)
- * handle this and read in the font when
- * appropriate. (Proposing ftstat() here).
- * Fix the ``smnt info''.