Re: problems on inline input on Mac

[YAMAMOTO Mitsuharu]

>>>>> On Fri, 28 Jul 2006 12:58:20 +0900 (JST), Kazu Yamamoto (山本和彦) 
>>>>> <address@hidden> said:

> Other inline input mechanisms (like "tamago") makes use of buffer
> itself. I guess many people agree that this approach is good.

It would be helpful to think the TSM support (including inline input
support) on Mac corresponds to the XIM support on X11 rather than
Emacs-level input methods such as tamago and leim.  Then I think
you'll naturally understand its behavior.

> Unfortunately, the inline input of Mac makes use of overlay instead
> of buffer. This approach causes some problems:

> * Even if the target buffer is read-only, we can input Hiragana and
>   Kanji. Read-only errors occur just after we complete strings (ie
>   inserting the completed strings from the overlay to the buffer).

The use of overlay strings is intentional and has some reasons.

  * Since TSM support corresponds to XIM, it is natural not to change
    the buffer contents until the user confirms the text.

  * Some keyboard layouts require to show unconfirmed (preedit) text
    even when we input some ASCII characters such as "`".  There may
    be some key binding including "`" that is also allowed to operate
    on read-only buffers.

Moreover, you cannot tell in general not only whether the target
buffer is read-only but also which is the target buffer at the stage
of passing input events to TSM using SendEventToEventTarget in
XTread_socket.

> * Since Hiragana and Kanji are hold in an overlay, text properties
>   of the buffer is not inherited during input. Of course, after
>   completing strings, the properties are inherited by the strings.

They are inherited from those just after the (empty) overlay.  You
cannot tell whether the user wants to inherit them from those before
or after it.

> So, I would like to ask to change the implementation of inline input
> on Mac from using overlay to using direct buffer.

I've made some email exchanges with Hashimoto-san (the author of
"inline-patch") on the above issues in Japanese.  I will send you some
of them if he agrees.

                                     YAMAMOTO Mitsuharu
                                address@hidden