Re: \bookpaper parameters ignored

[Werner LEMBERG]

> > Hmm, you've apparently used EC.enc for the font conversion.
>
> Yes, why is that Hmm?

`Hmm' means `thinking loud'. :-)

> > This uses glyph name `\hyphen' for position 0x2D, while lilypond's
> > latin1.enc uses `\minus'.
>
> Yes, I think ISOLatin1 says 0x2D is MINUS, not HYPHEN, and EC is
> wrong/different here?

The Adobe glyph list (AGL, which is the right authority if you handle
glyph names) says that `hyphen' is U+002D while `minus' is U+2212.
The Unicode standard calls U+002D `HYPHEN-MINUS'.

So EC.enc isn't wrong.  The intention of this glyph is clear, it is to
be used as a hyphen, *not* as a minus sign, which has a different
length: in TeX, you should e.g. write `$-1$', not `-1' to get a minus
glyph from a mathematical font.

Here you can see the typical problems if you want to make input
encoding exactly the same as the font encoding without a step
inbetween (namely, to use glyph entities).  I currently don't have a
listing of Adobe's ISOLatin1 font encoding, but it seems that EC.enc
is not a superset.

> I reluctantly changed this in latin1.enc, because/but I assume this
> will break other fonts that do not have this encoding quirk.

Having a single font encoding for all fonts doesn't make sense at all.
As the name already suggests, a font encoding is specific to a font
(family).  If you want to have a common font encoding used by really
many fonts, use Adobe's predefined font encoding called
`StandardEncoding'.  Unfortunately, this is a least common denominator
which means that the really interesting glyphs in a font are left
unencoded.  It is always expected that an application uses a proper
encoding vector to reencode the font appropriately.

> What I do not understand, is why EC.enc does not define MINUS at all
> eg, at an alternative postion.

Because EC fonts simply don't contain a MINUS.  Additionally, a PS
encoding vector has exactly 256 elements, and for EC fonts there isn't
a free slot which you could use for an alternative name.

I just can repeat that the font encoding model used in LilyPond is too
simple for real life.


    Werner