Re: bad afm file from Linotype

[Berthold Horn <bkph@ai.mit.edu>]

At 02:19 1999-02-05 +0100, Walter Schmidt wrote:

>I have downloaded a free Type1 font from the Linotype site 
><http://www.linotypelibrary.com>.  They are distributing one 
>variant (regular 430) of their new "Univers basic" font family.  
>They call it a "Demo" but they say it's not "crippled" in any way.

The errors in the AFM file are reflected in the actual font program itself.
This font has apparently not been fed through their QA department!

>Bad news is, that fontinst did not like the afm file when tried 
>to install it, so I checked the file contents and found very 
>strange things, e.g. the following line:

>CC idieresis 2 ; PCC  0 0 ; PCC dieresis -8 0 ;
>which should IMO rather be
>CC idieresis 2 ; PCC dotlessi 0 0 ; PCC dieresis -8 0 ;

Right, although, there is no reason for any software to look at these lines at all.
They tell you how the composite characters can be made by superimposing
base and accent.  But this isn't something you would ever want to do if
you have the ready-made composites.  Is it AFM2TFM that complains?

>I corrected several (!) bugs of this kind manually, and fontinst 
>ran smoothly then;  my bug fixes had, obviously, made some sense.
>The resulting font was, however, lacking the glypghs igrave, 
>iacute, icircumflex and idieresis  (only the accenty were printed).

The font in fact does not have the composites.  Or rather the code
for those composites is essentially the same as that of the
corresponding accents (except for the advance width).  One
can conjecture that the tool they use to make these fonts read the
same bogus entries in the AFM file and instead of stopping and
complaining just left out the base character.

There are more such errors, Zcaron, zcaron, Scaron and scaron
lack the caron accent --- which reflects the fact that the entries in
the Composites section of the AFM file lack the "caron".

>I thought about sending a complaint to Linotype, but I just
>do not know the technical background well enough:

>May I really assume that any "Type 1 for PC" font can be used 
>with TeX?

Yes.

>Do other (non-TeX) users not require afm files, so that nobody 
>cares?

Mostly they depend on PFM files in Windows or FOND resources
on the Mac.  But the bug is in the actual character outlines, so...

>I cannot imagine that Linotype is distributing garbage;
>perhaps fontinst isn't clever enough? 

In this case Linotype is distributing a font that is not free of errors.  
Also the hinting also is not good.  No hint switching for a start.

Regadrs, Berthold.



Berthold K.P. Horn	Cambridge, Massachusetts (M)