4 Mar
2018
4 Mar
'18
8:23 p.m.
I hadn't realized that groff hyphenation had been taken from
Tex, not troff. Is that becuase Tex did a better job, or
because troff's was deemed proprietary?
4 Mar
4 Mar
8:42 p.m.
New subject: [groff] The hyphenation algorithm produces wrong results
On Sun, Mar 4, 2018 at 3:23 PM, Doug McIlroy <doug(a)cs.dartmouth.edu> wrote:
I hadn't realized that groff hyphenation had been taken from
Tex, not troff. Is that becuase Tex did a better job, or
because troff's was deemed proprietary?
Given the author, I would guess the later as he wanted to be FOSS and
would not
have looked at the ditroff source - but that guess is worth just
that ;-)
ᐧ
9 p.m.
New subject: [groff] The hyphenation algorithm produces wrong results
On Mar 4, 2018, at 12:42 PM, Clem Cole
<clemc(a)ccc.com> wrote:
I remembered reading about Knuth's line-breaking algorithm in
Software Practice & Experience in early eighties and being quite
impressed with it. So may be that clear description of the algorithm
has something to do with it? Ah, here it is:
“Breaking Paragraphs into lines” by Donald Knuth & Plass,
SP&E, Volume 11, issue 11, Nov. 1981
(Download from Wiley is not free)
On Sun, Mar 4, 2018 at 3:23 PM, Doug McIlroy
<doug(a)cs.dartmouth.edu> wrote:
I hadn't realized that groff hyphenation had been taken from
Tex, not troff. Is that becuase Tex did a better job, or
because troff's was deemed proprietary?
Given the author, I would guess the later as he wanted to be FOSS and would not have
looked at the ditroff source - but that guess is worth just that ;-) 
9:32 p.m.
New subject: [groff] The hyphenation algorithm produces wrong results
On 2018-03-04 4:00 PM, Bakul Shah wrote:
On Mar 4, 2018, at 12:42 PM, Clem Cole <clemc(a)ccc.com
<mailto:clemc@ccc.com>> wrote:
That's the line breaker, which is an important contributor to the
quality of TeX output.
But TeX's *hyphenation* algorithm per se was invented by Franklin Mark
Liang and was indeed considerably better than its predecessors and
competitors (including most or all commercial typesetting software --
which was a big part of the motivation for it):
https://tug.org/docs/liang/liang-thesis.pdf
--Toby
On Sun, Mar 4, 2018 at 3:23 PM, Doug McIlroy <doug(a)cs.dartmouth.edu
<mailto:doug@cs.dartmouth.edu>> wrote:
I hadn't realized that groff hyphenation had been taken from
Tex, not troff. Is that becuase Tex did a better job, or
because troff's was deemed proprietary?
Given the author, I would guess the later as he wanted to be FOSS and
would not have looked at the ditroff source - but that guess is worth
just that ;-)
I remembered reading about Knuth's line-breaking algorithm in
Software Practice & Experience in early eighties and being quite
impressed with it. So may be that clear description of the algorithm
has something to do with it? Ah, here it is:
“Breaking Paragraphs into lines” by Donald Knuth & Plass,
SP&E, Volume 11, issue 11, Nov. 1981
(Download from Wiley is not free)
9:50 p.m.
New subject: [groff] The hyphenation algorithm produces wrong results
Hi Doug,
Bakul wrote:
I remembered reading about Knuth's line-breaking
algorithm in Software
Practice & Experience in early eighties and being quite impressed with
it. So may be that clear description of the algorithm has something
to do with it? Ah, here it is:
“Breaking Paragraphs into lines” by Donald Knuth & Plass, SP&E, Volume
11, issue 11, Nov. 1981
That's more to do with TeX looking at the whole paragraph when deciding
where to split lines. Hyphenation is part of that because a word might
help out by being the ideal thing to split and have the rest of the
lines sit easily in their length, but TeX's hyphenation algorithm is
distinct again.
Ted Harding gives some background on the groff list back in 2001,
https://lists.gnu.org/archive/html/groff/2001-03/msg00026.html
but I expect groff used TeX's algorithm because it was published, could
handle multiple languages, e.g. hyphen.us, and the data files were
available to contort into what groff ended up using in its simplified
TeX algorithm.
$ cd /usr/share/groff/1.22.3/tmac
$ ls hyphen*
hyphen.den hyphenex.cs hyphenex.us hyphen.sv hyphen.us
hyphen.cs hyphen.det hyphenex.de hyphen.fr
$
They've comments explaining their content.
Werner Lemburg on the groff list probably knows for certain as he had to
fathom all this out before becoming groff's excellent maintainer for
many years.
--
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy
10:36 p.m.
New subject: [groff] The hyphenation algorithm produces wrong results
On Mar 4, 2018, at 1:50 PM, Ralph Corderoy
<ralph(a)inputplus.co.uk> wrote:
Hi Doug,
Bakul wrote:
Thanks Ralph and Toby. “Because it was clearly described and published”
was the point I was trying to make and should’ve stopped there : ). SP&E
article had made a strong impression on me and that is what I instantly
thought of.
I remembered reading about Knuth's
line-breaking algorithm in Software
Practice & Experience in early eighties and being quite impressed with
it. So may be that clear description of the algorithm has something
to do with it? Ah, here it is:
“Breaking Paragraphs into lines” by Donald Knuth & Plass, SP&E, Volume
11, issue 11, Nov. 1981
That's more to do with TeX looking at the whole paragraph when deciding
where to split lines. Hyphenation is part of that because a word might
help out by being the ideal thing to split and have the rest of the
lines sit easily in their length, but TeX's hyphenation algorithm is
distinct again.
Ted Harding gives some background on the groff list back in 2001,
https://lists.gnu.org/archive/html/groff/2001-03/msg00026.html
but I expect groff used TeX's algorithm because it was published, could
handle multiple languages, e.g. hyphen.us, and the data files were
available to contort into what groff ended up using in its simplified
TeX algorithm.
$ cd /usr/share/groff/1.22.3/tmac
$ ls hyphen*
hyphen.den hyphenex.cs hyphenex.us hyphen.sv hyphen.us
hyphen.cs hyphen.det hyphenex.de hyphen.fr
$
They've comments explaining their content.
Werner Lemburg on the groff list probably knows for certain as he had to
fathom all this out before becoming groff's excellent maintainer for
many years.
--
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy
2251
days inactive
2251
days old
5 comments
5 participants
participants (5)
-
Bakul Shah -
Clem Cole -
Doug McIlroy -
Ralph Corderoy -
Toby Thain