[texhax] Hyphenation questions

Barbara Beeton bnb at ams.org
Thu Feb 2 23:08:14 CET 2012


doug mckenna asks:

    Question #1

    I found an English word that TeX (TeXLive 2010, running LaTeX under 
    TeXShop on a Mac) wasn't hyphenating correctly (the word was 
    "swimmingly").  But since TeX goes to some lengths not to hyphenate words 
    during line-breaking of paragraphs, I found it hard to isolate and test 
    what was going on.  As opposed to punting by simply adding a 
    \hyphenate{swim-ming-ly} exception declaration to my document, which of 
    course worked to solve my immediate problem, but didn't tell me much 
    about why the word wasn't otherwise being hyphenated right.

    Is there a simple macro or testbed file of TeX source code that allows 
    one to test any specific word (English or otherwise) to see how TeX/LaTeX 
    would hyphenate it in various circumstances?

simple.  using plain tex from the command line:

  tex
  **\relax
  *\showhyphens{swimmingly}

and yes, that is pretty bad.

i collect bad hyphenations, and publish
addenda to the list periodically in tugboat.
(an installment should appear in the next
issue.)  and whenever an addendum appears
in tugboat, the entire accumulation is
posted to ctan:

  http://www.ctan.com/info/digests/tugboat/hyphenex/

(swimmingly is already on the list; has
been for years.)  feel free to send me any
new exceptions that you stumble on, after
checking the existing list.

    Question #2

    So being of a curious mindset, I tried to understand \patterns, which 
    admittedly took me off on a tangent.  It seems that when hyphenation 
    patterns are installed before typesetting begins, duplicates (if any) are 
    complained about.  But what constitutes a duplicate?  Are the patterns 
    "mm1" and "mm3" duplicates?  "mm" and "mm1"?  "mm2" and "mm4"?  I ask 
    because there's some tests for duplicate \patterns in "trip.tex" that I 
    don't quite understand.  It seems to me that for any given sequence of 
    characters, whatever hyphenation "counts" (or whatever we call the digits 
    in a pattern) there are shouldn't affect whether the pattern is a 
    duplicate or not.

i disclaim any ability to unwind this
thread without serious research, for
which i haven't got time.
						-- bb


More information about the texhax mailing list