Re: [Eliot-dev] Catalan language file and dictionary

[Olivier Teuliere]

Hi Joan,

[I re-added the eliot-dev ML in Cc:, forgotten in my first reply].

The ca.po file is now in the repository, and will be part of the next
release. Thanks for your contribution!

Concerning the dictionary, you are right that digraphs are not
supported. What is supported, however, is the handling of any (single)
unicode character. This means that with the current version of Eliot,
you can already do much better than K, Q, W and Y. You could use for
example the following substitutions:
Ç-->Ç (no substitution needed!)
L·L-->Ŀ (U+013F)
NY-->Ñ (U+00D1)
QU-->Q
These are only suggestions, feel free to use more convenient ones. But
you shouldn't choose too exotic characters, because they will be hard
to input by the players (except with copy-paste)...

I won't save the current version of the dictionary in the repository,
because I prefer to store the uncompressed word list rather than a
binary file, even if .dawg files are offered for download on the
website. Could you please send me a link to the word list you used?

I agree with your analysis concerning digraph (or multi-characters)
support, but I need to know a bit more about Catalan Scrabble first:
1) Is it guaranteed that NY in a word is always played with one tile?
Or could it mean "N tile followed by Y tile"? Same question for QU.
2) How would players input "L·L"? It may not be easy to support
various input methods (like 'L', '.', 'L' being equal to 'L', '·',
'L'), so I'd prefer to be sure each tile has only one "canonical"
input method.

Best regards,
-- 
Olivier

On Mon, Jun 8, 2009 at 12:54 PM, Joan Montané<address@hidden> wrote:
> Hi Olivier,
>
> please, reply when finish your holidays. Thanks.
>
>
> I'm unable to create dawg file with compdic. It shows an error at
> multiletter tiles. In Catalan case:
> L·L 10  1 0 1
> NY  10  1 0 1
> QU  8  1 0 1
>
> Currently, this lines are replaced, also in dictionary file, by free
> letters, all works fine then:
> W 10  1 0 1
> Y  10  1 0 1
> Q  8  1 0 1
>
>
> I think the good aproach about multiletter tiles is:
>
> 1.- Internally, Eliot must works always with the multi-letter tiles
> replaced by a single free letter.
> 2.- All the user input should be parsed to internal mode (E.g.: user
> inputs PARAL·LEL and Eliot parses the input to PARAWEL, and works
> internally with PARAWEL)
> 3.- All the output lists done by Eliot are parsed again (i.e: lists
> are showed in "human mode": PARAL·LEL)
>
> To achieve this, dawg format must be extended with and optional 6th
> column for letters definition file, something like this
>
> W 10  1 0 1 L·L
> Y  10  1 0 1 NY
> Q  8  1 0 1 QU
>
> Then, Eliot uses this info to parse the user's input (L·L-->W) and
> Eliot's output (W-->L·L).
>
> What do you think about?
>
> Joan Montané.
>
> 2009/6/4 Joan Montané <address@hidden>:
>> Hi Olivier,
>>
>> I will remake and check an Eliot dictionary with digraphs (L·L, NY,
>> QU). May be I've make some mistake.
>>
>> I have some requests, specially for duplicate mode. I will send them
>> in two weeks. I want to do a good test and look the code (I'm not a
>> programmer-guy, but I can "understand" code a little). I think Eliot
>> could the best manager for Duplicate games.
>>
>> Thanks for your work and enjoy holidays! :-)
>>
>> Joan Montané
>>
>> 2009/6/4 Olivier Teuliere <address@hidden>:
>>> ¡Hola Joan!
>>>
>>> On Wed, Jun 3, 2009 at 10:14 PM, Joan Montané <address@hidden> wrote:
>>>> In [1], you can download a ca.po file with (almost) all the Eliot strings
>>>> translated, and in [2] a Catalan dictionary file, based on a good
>>>> enough train wordlist, not official, neither perfect, but the best
>>>> publicy avaiable in Catalan. Please, add them to the trunk code,
>>>> thanks
>>>
>>> Thanks, I will do it in about 2 weeks... when I am back from my
>>> holidays in Spain :-)
>>>
>>>> There are at least three languages using digraphs on Scrabble:
>>>> Catalan, Spanish, Welsh. So, I request support for digraphs tiles on
>>>> Eliot project.
>>>
>>> Since several versions now, Eliot is supposed to support any set of
>>> characters (up to 63 characters). I once tried to build a dictionary
>>> with Polish words, and both the dawg creation and the display on the
>>> interface were fine, so I see no reason why Catalan would not work.
>>>
>>> There is one condition though: the input files for the 'compdic'
>>> program should be encoded in UTF-8. Was it the case when you tried?
>>>
>>> If there is really a bug, I will also solve it when I am back home.
>>>
>>>
>>> Thanks for your contribution!
>>>
>>> Best regards,
>>> --
>>> Olivier