Re: Some points about IM integration

[Sheng Mao <sherlockmao gmail com>]

On 14/05/2012 22:52, Germán Póo-Caamaño wrote:
> On Tue, 2012-05-15 at 12:21 +0800, Weng Xuetian wrote:
> > > I don't want people to draw the conclusion that because I'm saying
> > that
> > > input methods should have simple configuration without a lot of
> > options,
> > > I think that they aren't important. I'm very aware that every single
> > > user that comes to GNOME and wants to write in Chinese needs to use
> > an
> > > input method. But if we have so many options that the defaults don't
> > get
> > > well tested, or if options conflict and produce bugs, then we're not
> > > shipping a good ';';'''''''
> > > '
> > Options are really required in order to meet people need especially
> > for input method. This is a basic components for people.
>
> Is it possible you can enumerate all those special needs and why are
> compulsory?
>
> Just stating the options are important does not help to understand why,
> neither gives the opportunity to think or determine which approach would
> fit better.

These are common options for Chinese IMs:

    * Dialect-specified pronunciation: There are nine major dialect 
families for Chinese, some of which are different from each other more 
than Dutch and German. The problem is some pronunciations cannot be 
spoken by some dialect users. Thus, they need special settings to enable 
other pronunciations to replace the unspoken ones.

    * Words-candidate-list: Each Chinese character has a one-syllable 
pronunciation, and a Chinese word is made up by 2~4 characters 
(generally speaking), which correspondingly has 2~4 syllables. When a 
N-syllable pronunciation is input, IM engine should enumerate all 
possible combinations, while it may be a new word and the user should 
pick characters one by one to make up the new word. The question is how 
many candidates will be shown to users before he/she can choose single 
characters.

These are two of the most popular options for nearly all Chinese IMs. 
They are not merely options, but rather _fundamental_ features.

I think one framework for all input method is ideal, and multiple 
upstream packages would cause weird bugs, and users are prone to blame 
Gnome for all bugs. Also, I know Gnome team needs an IM team to work 
with closely and ibus team just fulfill your needs well, I understand 
that you need some one to be the fire rescue when there are bugs.

I think the discussions from Weng and Su show two ideas:

    * As features, fcitx, with plugin mechanism, is more powerful than ibus
    * As philosophy, they hope Gnome won't expel them out.

Maybe you can choose a default framework for Gnome, and it is your 
project and you have the right to make decision, but is there any chance 
*not* to impede the existence of other IMs.

@Weng Xuetian

Thank you for developing fcitx!

I am wondering, if any detailed plan on how to cooperate with Gnome IM 
design will be useful. I think they need some commitment.

> _______________________________________________
> desktop-devel-list mailing list
> desktop-devel-list gnome org
> http://mail.gnome.org/mailman/listinfo/desktop-devel-list