Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site sunybcs.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!rochester!rocksanne!sunybcs!colonel
From: colonel@sunybcs.UUCP (Col. G. L. Sicherman)
Newsgroups: net.text,net.bugs.usg
Subject: bug in makedev.c (ditroff support)
Message-ID: <2399@sunybcs.UUCP>
Date: Wed, 16-Oct-85 11:21:07 EDT
Article-I.D.: sunybcs.2399
Posted: Wed Oct 16 11:21:07 1985
Date-Received: Sat, 19-Oct-85 04:46:18 EDT
Distribution: net
Organization: Save the Dodoes Foundation
Lines: 28
Xref: watmath net.text:678 net.bugs.usg:365

["At last!  My new gorilla-detector machine!  They can't fool me now!"]

There's an odd and significant error in how makedev computes the wordspace
size for a font.

In troff, the word spacing is measured in thirty-sixths of an em (why?),
and the default is 12.  The .ss directive alters this value.

In ditroff, the word spacing is also stored in the font tables, which
produces a conflict.  Ditroff uses the table value, and interprets .ss
by adjusting the space proportionately to the initial .ss value of 12.

The makedev default is also 12 thirty-sixths (one-third) of an em.  But
where ditroff always rounds its divisions, makedev truncates this one.
This discrepancy won't mess up your output, but it may make the
wordspaces a little narrower than they ought to be.

The fix, of course, is to replace makedev.c's truncated division by a
rounded division.  Since the source is controlled, I won't post any of
it; you should have no trouble finding the offending statement.

The quick (?) fix is to specify "spacewidth" explicitly in all your
font tables.
-- 
Col. G. L. Sicherman
UU: ...{rocksvax|decvax}!sunybcs!colonel
CS: colonel@buffalo-cs
BI: csdsicher@sunyabva