[tex-live] patterns generation & editing (was: your eqexam package)

[cfrees at imapmail.org]

On Sat 10th Jul, 2010 at 01:17, Manuel Pégourié-Gonnard seems to have written:

> Hi Taco,
>
> First of all, I'd like to make clear I have no opinion on what is the source of
> a pattern file, so the question below is no an argument in the "source"
> discussion, but really pure curiosity.
>
> Le 09/07/2010 22:43, taco a écrit :
>> Philip Taylor (Webmaster, Ret'd) wrote:
>>>
>>> cfrees at imapmail.org wrote:
>>>
>>>> But if it is the preferred form for editing and 'source' is, in this
>>>> context, defined to be the preferred form for editing, then it is the
>>>> source in the relevant sense. Of course, if it is *not* the preferred
>>>> form for editing or if 'source' is defined differently, then it is not.
>>>
>>> I think I can say with 99.9% confidence that it is not the
>>> preferred source for editing.
>>
>> I'm willing to raise the confidence level by a couple of nines
>> on that. Editing patgen-generated files to fix hyphenation issues
>> is next to impossible, you really need the source word lists
>> (unless you have an full and correctly hyphenated vocabulary of your
>> language in your head and are able to juggle all the juxtapositions
>> of all the hyphenation points in them without making errors or omissions).
>>
> I'm quite surprised to learn that. I know near to nothing about the process of
> building pattern files, but, being French, I already had a look at hyph-fr.tex,
> and noticed it is full of comments, and moreover the patterns are divided as
> "phonetic" and "etymological". Here is an excerpt:
>
> %-------------------%-----------------------%
> % phonetic patterns % etymological patterns %
> %-------------------%-----------------------%
> %%a
> .a4
> 'a4
> .â4
> 'â4
> ab2h % df-bg 1998/02/07 for abhorrer
>                    .ab3réa
>                    'ab3réa
> ad2h % df-bg 1998/02/07 for adhèsion & co
>                    a1è2dre
>                    .ae3s4ch
>                    'ae3s4ch
>                    1alcool
>
> It really give the impression the pattern list was edited by hand, sometimes
> with the intention of fixing hyphenation for a particular word, which looks
> quite contradictory to your remark.
>
> Could it be that it depends very much on the language, and that French would be
> a particularly regular language wrt hyphenation? Or am I misinterpreting the
> presence of the comments?

Looking at a couple of files was really what prompted my earlier
comments. I don't know anything about doing this. The file for Welsh,
for example, doesn't have the sorts of comments you see in the French
file but the commentary at the top seems to assume that people would
edit the file directly. For example:
% This file is part of hyph-utf8 package and resulted from
% semi-manual conversions of hyphenation patterns into UTF-8 in June
% 2008.
%%
% Source: cyhyph.tex ()
% Author: Yannis Haralambous
%% The above mentioned file should become obsolete, 
% and the author of the original file should preferaby modify this file instead.
etc. etc.

Obviously the same isn't true for all languages - that's what happens
when I peak without finding out properly!

- cfr