Re: 3.6 Feature: IBus/XKB integration

[Marguerite Su <i marguerite su>]

Hiya, all,

I followed up this thread because I saw Vincent continued our discussion here.

At first I thought it might be enough, Vincent is a core person can
fully standing for a distribution's gnome community.

now I'm on behalf of Chinese community to add some background knowledge here.

if you follow our distro thread too, you may find I have a survey. and
luckily, it generates some results.

to start up, I must say sorry. because most of Chinese are not good at
English, so they make you GNOME people really hard to hear any
feedback from CJK community (C stands for Chinese, why it comes first,
because it has the most large potential user group)

and the input method situation in C and J community is (lack of K
community's feedback):

1. scim is dead in development. weng and me both know its father, he's
acting as upstream of input method, that is, the "method" like how to
split word from sentences(it's easy in English and western language,
but really hard in CJK.)

2. ibus is a complicated issue. the largest three distribution in C is
Ubuntu, Fedora, and openSUSE.

    Ubuntu and openSUSE have fcitx and ibus at the same time. they're
the target of my suvery. that is: if user has choices, what his input
method will be?

    Fedora is anther story. you may know fedora uses Bugzilla as
package review center too. fcitx packages are in waiting state for
almost two years. why? because its developers are "developing" ibus if
you can call it developing (in my point of view, they're making no
progress but maintenance.), and they don't know (or don't care to
know) what fcitx actually is, like you.

    Actually fcitx is not something new toy. if it is, Vincent will
never help us like this, because it's not mature enough to replace
ibus as the default method of a so large distribution. fcitx is being
developed since 2004. in CS, 8 years is enough for a babe to grow up.

    so Fedora/Red Hat is not my target. it's not because I don't want
to, but because they're reacting so slow that even can't meet my
basis: you should have both ibus and fcitx in your distro.

the temporary result of my survey is:

( I posted my survey on G+, Facebook, forums. it has enough receivers.)

top choice for zh_TW: gcin and its derivatives hime. (100%)

top choice for zh_CN: fcitx (100%)

there's no even a single person choose ibus if he has choice.

and what more interesting is, some users( 2 or 3) is now using ibus,
but they vote for fcitx/gcin. why? because ibus is default shipped.

and still what even more interesting, Japanese community calls for
development of fcitx-anthy. but they have already had ibus nowadays.
if you're living in a house luxurious enough, do you still need
another house to live?

that's pretty enough. I think.

but like I said in our distro thread, we can't hold you back from
"integrating" ibus into GNOME. reasons:

1. LOSS spirit is developer can do anything he is interested.

2. Fedora ( see reasons above) still has no official alternatives.

just when you're developing, please keep this in mind that:

 there certainly are "other input methods" existing and actually
they're nowadays user preference.

so no need to integrate ibus that tight, leaving no space for other
input methods to perform and survive.

actually "what you think" is not so important, because end users will
vote by foot. (but "what you need" is really important, because you're
still one of the end users, the only different is that you can
develop)

if in GNOME 3.6, I install fcitx on my laptop, but there're still
libibus library or something left, because GNOME needs it, then it
becomes an critical case:

one system, two input methods, one can't be fully removed. are we on
Windows again? are we GNU?

then what can I do? to uninstall GNOME? that's not a good choice. I
may have to endure it.

then it might be not possible for any distribution contains gnome to
change their default input method.

because they certainly don't want to acting fool. leaving something
can't be uninstalled. users will beat them to death in forums.


so pretty clear now?

1. ibus is not the first user choice if he has a choice.

2. leave something can't be uninstalled while user does not need it is not GNU.

3. in this case, you're really misrepresenting end users. you're
planning to give them something they don't need so eager and most of
them don't even need it.

   actually you're doing some backport work, for fedora only. so we
can't hold you back. but you can't push that requirement to all users,
including ones who have already has a better choice.

all the things I said here is based on facts, not theory. so please do
listen to it.

LOSS developers do can invent anything interested in, but no users or
what's worse, users leave because of it (I say that because it has
already happened, one of my friends left ibus to embrace fcitx,
because he learned ibus might be compulsory. LOSS users really hate
that word) then it doesn't make any sense of the development and will
truly break the developer's heart.

it sounds to me that to invent stream engine when there's already
electricity everywhere.

sad story.

hope it helps

Marguerite