Re: 8r fonts

[Constantin Kahn <kahn@math.uni-hannover.de>]

Sebastian Rahtz writes:
>bleeargh. but do i leave 8r.mtx as it is? Constantin, can you look?

I tend to think that Thierry (and other supporters of his views) are right.
However, there haven't been any text samples for others to view, there
hasn't been a comprehensive discussion of the kerning pairs of any single
font (not even Times for which only selected pairs of characters were
quoted), and there hasn't been any discussion of how this issue affects
fonts from different vendors (other than references to URW fonts).

While I find the underlying argument *very* plausible I must say that I
find the data to back up the claim utterly unconvincing. (E.g., I believe
that I could find poorly kerned pairs in practically every font if I look
long enough; if URW fonts are noticeably affected, then it's only more
important to check whether other vendor's fonts suffer from the same
problems; finally, I don't think that having a lot of kerning pairs is
necessarily an indication of quality for a font: it could as well be an
indication of sloppy design- not to slighten URW's offerings which I don't
have any experience with!) Unfortunately, I don't have the time now and I
don't have a font archive of sufficient breadth to "look", sorry.

On the other hand, I think that font metrics from a single source (in this
case: Sebastian) should be consistent across encodings. In particular,
kerning pairs shouldn't change when changing encodings (as long as a
character pair is present in both encodings). The old 8r.mtx (mine)
violated this principle (for reasons which I've tried to explain in an
earlier post), the new one seems to adhere to it. As long as Alan doesn't
intend to change latin.mtx the new 8r.mtx is better and the old one should
be considered a mistake.

So I am with Thierry, only for different reasons (while sympathizing with
his views). Is this good enough?

Besides, despite the fact that it has my name on it I do not demand custody
of 8r.mtx. I've always considered it a one-time contribution (and I
remember that I was surprised when it found its way into psfonts
unchallenged and unchanged).

Constantin

--
Constantin Kahn, Am Fuhrenkampe 80, 30419 Hannover, Germany
Phone: (+49) 511-762-5380 or (+49) 511-758528 (home)
Email: kahn@math.uni-hannover.de