Re: [m17n-list] Arabic translit IME

[Joop Kiefte]

Thanks for all the great comments! I think most of the suggestions you provide are great, and if you need me to agree to apply those, consider it granted. I don't speak Arabic myself yet, so of course I have probably made some glaring omissions... :(

Anyway, with a first version submitted, it becomes easier to send suggestions as patches instead of (generally interpretable in several ways) general suggestions, so let's not wait before all comments are received.

On Fri, Aug 15, 2014 at 5:16 PM, Iori　YONEJI <address@hidden> wrote:

assalam again,

I've posted about your patch on twitter, I've got some mention about this.
(sorry we talked in Japanese, but Naoto and Kenichi can read, I think ;p)

https://twitter.com/na4zagin3/status/500283008662790144
https://twitter.com/na4zagin3/status/500283881887506432
https://twitter.com/na4zagin3/status/500286795754377216

Summarily, he advised me that you forgot kasratan, and because it's
tend to be used often, so it's better to able to type this easier.
Also, fatḥatan and ḍammatan is important too. With less importance,
but there still some characters waṣla and tatweel.

So, I tought that:
1. These notation has particular sound (in, an, un) and indicate case
of word, so I think it should be <vin van vun>, <cin, can, cun> or
<xin, xan, xun>.
2. With the rule above, sequences of kasrah, fathah and dammah should
be representing their own sound. For example, <ci, ca, cu>. (By the
way, most cases of Arabic text writing, these notation may be
omitted.)
3. Expanding 1 and 2 rule above, representing sukun with single 'c' is
natural. Also, shaddah should be represented with "cc" as an example.
4. Tatweel is somewhat like underscore or dash, but can be connected
with neighbor characters. So examples are <c->, <c_>, <x->.....

I dont' have good idea about other stuff he and wikipedia
(http://en.wikipedia.org/wiki/Arabic_diacritics) mentioning yet..

On Fri, Aug 15, 2014 at 10:20 PM, Iori　YONEJI <address@hidden> wrote:
> Hi,
>
> I was impressed with your mim file that enables typing Arabic like as
> typing Latin characters.
> But I found some unpleasing sequences.
> How about minor -updating these points below:
>
> 1. ة is always placed on last character of a word, so the sequence
> <h.> is more preferable than <.h>. With this change, we can terminate
> femininum word with '.'.
> 2. ح is now <.7>, but how ِis <.h> or <Shift+h> ? This character (and
> more characters like ط or خ) is used often but typing <.7> is too
> hard, and they are not related with.
> 3. Mainly, Arabic scripts use ‏ «…» as quotation marks (probably same
> as French one). How do you think to add a rule to type them?
> http://en.wikipedia.org/wiki/International_variation_in_quotation_marks
> 4. There are more additional characters in Persian, Darī, and other
> languages. For example, گ (g), پ(p), چ(č) and ژ(ž). So I think it's
> better to preserve these key(g, p..). As another option, we can assign
> them as <k-> to گ (this character is a derivative of ک (k)) and پ as
> <p.> and so on.
> http://en.wikipedia.org/wiki/Persian_language#Additions
>
> Also I agree with Naoto's opinions.
>
> I attached a file as one of my attempt, but probably this isn't perfect one.
> Probably (".2" "ء") is also non-associative relation, and we left some
> keys that are on sweet location but not used (like 'c'), so there may
> be more needs to careful consideration.
>
> On Fri, Aug 15, 2014 at 1:45 PM, Naoto Takahashi
> <address@hidden> wrote:
>> On 14 August 2014 23:44, Joop Kiefte <address@hidden> wrote:
>>> Here a version now with vowels added.
>>
>> I have three points to mention.
>>
>> 1.  "a" is assigned to both ALEF U+0627 and ALEF MAKSURA U+0649.
>> Since the definition of ALEF comes first, there is no way to input
>> ALEF MAKSURA.
>>
>> 2. How about adding ".e", ".i", ".u" for FATHATAN U+064B, DAMMATAN
>> U+064C, KASRATAN U+064D, respectively?
>>
>> 3. Users would expect ARABIC COMMA U+060C and ARABIC SEMICOLON U+061B, too.
>>
>> --
>> TAKAHASHI Naoto
>>

--

Joop Kiefte

BM: BM-2cU7makjmuWMHXdjufjBqcSNLzyznTyk8E (web)