[Fontinst] Need help with Hoefler Text Type1

[Lars Hellström]

10 aug 2007 kl. 20.13 skrev Robin Schwab:

> Hi
>
> I'm trying to create a LaTeX adaption for Hoefler Text Type 1 on a
> Windows machine. This font comes with almost every variation you might 
> wish.
>
> The base encoding T1 contains hanging figures and the ligatures fi and
> fl. The extended encoding TS1 contains additional characters. I can
> access all those chars with my adaption produced with fontinst.
>
> However there are two big problems: 1st the expert set containing all
> ligatures ff, ffh, st, ... does not work. I think the reason for that 
> is
> that they are not in the .afm file. Any ideas where they could be?

In an expert font, perhaps? Classically vendors have organised their 
fonts so that there are no more than 256 glyphs in each, to make it 
possible for pre-Unicode software to access all the characters (typists 
were expected to explicitly switch to a different font when they needed 
characters not in the standard encoding, a scheme that was workable in 
WYSIWYG environments but not for LaTeX). Adobe Garamond is organised 
like this, and various examples of how to take advantage of (or leave 
out) its expert fonts are included in the fontinst distro as 
basicex.tex.

My guess would be: If you tell fontinst to use expert fonts, you will 
automatically get the ff, ffi, and ffl ligatures. In order to get more 
ligatures (e.g. ffh), provided they are present in the base font, you 
need an ETX which includes these; John gave some instructions on that. 
A catch is that since T1 already puts something in every slot you need 
to take some characters out in order to fit additional ligatures in. 
Primary candidates are the Dutch ligatures IJ and ij, but then things 
get harier.

> Second thing is that I get twice as much overfull boxes with hoefler
> text as with latin modern. It looks like the elasticity of the spaces 
> is
> wrong.

That's easy enough to check: what does the FONTDIMEN section in the 
generated VPL files say? You might also want to compare with the number 
of overfull boxes for Times; as I recall it, Computer Modern and 
derivatives have wider spaces than postscript fonts tend to have.

Another reason could be that hyphenation is turned off for this font 
(\hyphenchar set to -1), but I can't see why that would happen.

> I'm trying for weeks now and asked in some forums without success.
> You're my last hope to get the font working.

Well, it sounds like just our kind of question.

> Regards
>
> Robin
>
> PS: See the .afm file below for the missing ligatures.
>
> StartFontMetrics 2.0
> Comment Generated by PC Font Inspector 1.0.1 6/11/00 11:59:57 AM

AFM files that one generates oneself *sometimes* lack glyphs that are 
present in the font but the converter program considers unimportant 
(e.g. because it's not in the standard encoding!). Since this one 
claims to be seven years old however, it's more likely generated at the 
foundry.

[snip]
> C -1 ; WX 780 ; N infinity ; B 50 84 730 357 ;
> C -1 ; WX 406 ; N plusminus ; B 21 0 388 469 ;
> C -1 ; WX 442 ; N lessequal ; B 45 0 387 471 ;
> C -1 ; WX 442 ; N greaterequal ; B 58 0 400 471 ;
> C -1 ; WX 523 ; N mu ; B 1 -279 512 435 ;
> C -1 ; WX 518 ; N partialdiff ; B 35 -21 488 722 ;
> C -1 ; WX 669 ; N summation ; B 40 -10 623 694 ;
> C -1 ; WX 817 ; N product ; B 28 -6 789 690 ;
> C -1 ; WX 551 ; N pi ; B 15 -21 557 451 ;
> C -1 ; WX 260 ; N integral ; B -69 -278 330 722 ;
> C -1 ; WX 805 ; N Omega ; B 21 -6 788 705 ;
> C -1 ; WX 481 ; N logicalnot ; B 20 16 446 249 ;
> C -1 ; WX 474 ; N radical ; B 6 -22 440 719 ;
> C -1 ; WX 406 ; N approxequal ; B 31 97 383 349 ;
> C -1 ; WX 709 ; N Delta ; B 20 -6 689 705 ;
> C -1 ; WX 500 ; N nbspace ; B 0 0 0 0 ;
> C -1 ; WX 393 ; N divide ; B 20 58 373 383 ;
> C -1 ; WX 504 ; N lozenge ; B 45 -20 459 696 ;

The glyphs above are probably included to cover everything in the 
MacRoman encoding. If well-designed, they could be useful in TS1 or 
math encodings.

> C -1 ; WX 831 ; N Acircumflex ; B 35 -21 807 721 ;
> C -1 ; WX 780 ; N Aacute ; B 30 -5 780 690 ;
> C -1 ; WX 1164 ; N Iacute ; B 29 -264 1124 690 ;

That's a *very* odd width for an Iacute. It might be a different glyph 
which has been given an incorrect name to squeeze the font into a 
nonstandard encoding, so you might want to use some graphical font 
editor to check it out. It *could* be that this is where your missing 
ligatures are hiding.

> C -1 ; WX 461 ; N Idieresis ; B -175 -305 508 705 ;
> C -1 ; WX 326 ; N Oacute ; B 8 -54 300 434 ;
> C -1 ; WX 381 ; N Ocircumflex ; B 35 -13 349 457 ;
> C -1 ; WX 734 ; N apple ; B 45 -22 708 705 ;
> C -1 ; WX 369 ; N Ograve ; B 39 -13 336 420 ;
> C -1 ; WX 350 ; N Uacute ; B 107 -225 296 26 ;

Lars Hellström