PostScript rendering bug involving word with an umlaut

[Dave Kemper]

$ cat test.roff
.nr pp 14
.lp
coo\*:perate
.bl
.lp
blah blah blah blah blah
coo\*:perate
blah blah blah blah blah blah blah blah blah
.bl
.lp
blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah
coo\*:perate
blah blah blah blah
.bl
.lp
blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah
coo\*:perate
blah blah blah blah
$ groff -m me test.roff > test.ps

Examining the test.ps thus generated reveals:

 - In the first paragraph, the word with the umlaut is flush with the
 left margin, and renders correctly.

 - In the second paragraph, the word with the umlaut is in the middle
 of a line, and renders correctly.

 - In the fourth paragraph, the word with the umlaut is in the middle
 of a line where groff has inserted breaks, and renders correctly.

 - But in the third paragraph, the word with the umlaut does not render
 correctly when groff has inserted breaks that leave the word flush with
 the left margin.

This happens with the paragraphs in any order in the source file.  The key
to triggering the bug seems to be the combination of groff automatically
inserting breaks into a paragraph and the word with the umlaut appearing
first on a resulting line--note that the problem disappears if a ".brp"
is placed right before the umlauted word in the third paragraph.