Re: bug in ae virtual fonts, or in fontinst, or in vptovf/vftovp?

[Olaf Weber <olaf@infovore.xs4all.nl>]

Vladimir Volovich writes:
> "BKPH" == Berthold K P Horn writes:
>>> % vftovp aer10.vf aer10.tfm aer10.vpl Bad VF file: Oversize
>>> dimension has been reset to zero.

>  BKPH> I don't know whether this is the same issue, but many VF files
>  BKPH> insert absurd rules with semi-infinite negative dimensions.  I
>  BKPH> reported on this before, wondering whether this was some
>  BKPH> special marker used for some obscure purpose.  Of course, it
>  BKPH> doesn't print because the dimension is negative, but what is it
>  BKPH> for, and where does it come from?

> Note, that the original VPL file contained a zero-width rule (i.e. not
> with "semi-infinite negative dimensions"):

> (CHARACTER D 23 (COMMENT compwordmark)
>    (CHARWD R 0.0)
>    (CHARHT R 4.29993)
>    (CHARDP R 0.0)
>    (MAP
>       (SETRULE R 4.29993 R 0.0)
>       )
>    )

> So perhaps this shows a bug in vptovf? Or is vftovp incorrectly
> "interpreting" a zero width? Anyway, it looks like a bug in
> vptovf/vftovp, and not a bug in fontinst/ae fonts.

It looks like this was due to a bug in vftovp that was corrected in
the August 1998 release of vptovf (version 1.5) which had been found
by Wayne Sullivan.  This bug resulted in the semi-infinite negative
dimensions seen by Berthold.

The bug should be fixed in recent teTeX-0.9 snapshots, and will be
fixed in the upcoming web2c 7.3.


The following change summarizes the fix:

@x [128] Correct a bug found by Wayne G. Sullivan
if x>0 then negative:=false
@y
if x>=0 then negative:=false
@z

-- 
Olaf Weber