IBus UI review
by Matthias Clasen
Not sure if this is the correct place to send this, but I'll send it
here anyway (please tell me if there's a better place).
I've recently installed ibus in order to get some impression of how our
new im framework will integrate in the desktop. While playing with it, I
took some notes, that I'd like to share.
Matthias
---
Status icon
- The tooltip "IBus - Running" is pretty pointless less and should be
removed until there is something useful to say
- There is no way to switch back to "no input method" from the status
icon. I have to press Ctrl-space to go back. Maybe add an "None" entry
at the bottom of the menu ?
Toolbar
- Why do input methods seem to fancy these weird undecorated floating
toolbars ? Does it add anything that is not already present in the
status icon ?
- If we can't drop it, can there at least be a way to turn it off ?
- The toolbar seems useless if "focus-follows-mouse" is turned on, since
it becomes inactive on focus out. This also affects the status icon.
Menus
- What is the plan, going forward, wrt to im-chooser ? I'd hate to have
2 input method related menuitems in the default install. My preference
would be to not install the im-chooser by default, since it is only
needed to switch back to 'legacy' frameworks.
- It would be great if we could use the generic "Input Method" menuitem
for the ibus preferences, and maybe rename im-chooser to "Input Method
Framework" or something like that.
- Alternatively, if we can't get rid of im-chooser by default, maybe
ibus-setup should not have its own menu item (I notice that scim-setup
doesn't have one either), since it is available via im-chooser.
- There is a mismatch between the menuitem and the ibus-setup window,
both the window title and icon don't match the menu, as they should.
Preferences, General tab
- "Aauto start IBus on session login" is very techno babble. Can we make
that something like "Enable Input Methods" ? I don't think there is any
need to talk about sessions and autostart here.
- Keyboard shortcuts: I would love to see these moved to the keyboard
shortcuts capplet, which has support for handling application-defined
shortcuts. As a bonus, you get automatic conflict handling. The one
restriction is that currently, only one key-combination per action is
possible. If having multiple is essential, you could either split it
into "Trigger", "Alternative Trigger", "Second Alternative Trigger", or
file a bug and I'll look into enabling multiple shortcuts per action in
the keybinding capplet
- "UI" is a bad section label. How about "Fonts & Style" instead ? Even
better would be to split it into two sections, a la
Input Window
Lookup table orientation: [Vertical]
[ ] Use the system font
Input Window Font: [Sans 10]
Language Bar
[ ] Show language bar
[ ] Hide language bar when it is not needed
Preferences, Engine tab
- "Engine" is a technical term that is not really helpful here. How
about "Languages" instead ?
- There are some icons missing in the combo box, e.g Telugu-apple,
Telugu-rts, Marathi-phonetic, Marathi-itrans...
- The main list needs to repeat the language name (like the status icon
menu already does) otherwise it is not clear if "phonetic" is Oriya or
Marathi.
Preferences, About tab
This should not be done as a tab, it is very much against the style of
our preference tools. There is already an About menu item on the status
icon. If you absolutely want to have an "About" in the preferences, it
should be a left-aligned "About" button in the action area that brings
up an about dialog. But I'd really just get rid of it.
14 years, 11 months
How to convert Hanja with ibus?
by Warren Togami
https://bugzilla.redhat.com/show_bug.cgi?id=487269
This bug was closed because Hanja conversion was implemented in
ibus-hangul, but how are you supposed to use it? The regular hotkeys
Right-Alt and Right-CTRL seem to do nothing.
On the topic of hotkeys for Korean, is the Hangul/English button and
Hanja button in the positions of the Right-Alt and Right-CTRL
respectively the same keycodes with a different label?
Warren Togami
wtogami(a)redhat.com
15 years
fedora-i18n bug day 2 April
by Jens-Ulrik Petersen
The Fedora I18n Project is planning a bug day this Thursday (2nd April).
http://fedoraproject.org/wiki/I18N/Bugs has links to various lists of i18n related bugs.
The main focus will be triaging rawhide bugs related to i18n but also checking on older i18n bugs to see if they are still present in F10 and/or F11 Beta.
Join us on #fedora-i18n during the day to help out, ask questions, and for discussion i18n issues.
Fedora I18n Team
15 years
Announcing Transifex 0.5
by Dimitris Glezos
Indifex and the Transifex Community are proud to announce the newest version
of their flagship translation platform, Transifex 0.5.
Transifex is a web application written in Python using the Django web
framework that gives translators a web interface to various version control
systems. Files to be translated can be downloaded, translated files can be
uploaded directly to the source repository, and various translation statistics
can be read at a glance.
Transifex is already in use by the Fedora Project to translate its interfaces
to an audience of more than 5 million users.
What does it offer?
===================
Transifex currently supports the following Version control systems:
- Concurrent Version System
- Subversion
- Bazaar
- Mercurial
- Git
For statistics generation, Transifex supports static gettext message catalogs
and intltool-based ones, used by the vast majority of open source software
projects.
What's new in 0.5?
==================
A full list of the features offered in this release can be found in the
release notes:
http://docs.transifex.org/releases/0.5.html
This release represents a significant advance in Transifex development since
the Transifex 0.3 release in November 2008. Here’s a 40K-foot view of the
release in numbers and most important feature categories.
243 files changed, 14027 insertions(+), 319 deletions(-)
- Complete re-write of the source code on top of the Django Web framework
- New data model supporting multiple repositories per project (eg. branches
or domains of files), and project collections (eg. Fedora, GNOME, etc.)
- Calculation of a project’s translation coverage (statistics)
- Submission support of files to a variety of version control systems
- Support for serving translation files to users for easy access to them
- User registrations and authentication (including OpenID)
- Simple workflow support
What does it look like?
=======================
The Fedora Project is currently running an instance of Transifex at:
https://translate.fedoraproject.org/tx/
For some eye candy, check out the screenshots on our site, at:
http://transifex.org/screenshots
How can I get it?
=================
Project and community managers who want to deploy Transifex for their
own community can get Transifex in a variety of ways.
A tarball of Transifex 0.5 is available at:
http://transifex.org/files/
For full installation instructions, refer to the documentation section:
http://docs.transifex.org/intro/install.html
RPM packages for Fedora 9, Fedora 10, and Fedora Rawhide are or will
soon be available via yum:
yum install transifex transifex-extras
RPM packages for RHEL 5, CentOS 5, and Scientific Linux 5 will be
available in Fedora EPEL:
https://fedoraproject.org/wiki/EPEL
Regards,
The staff of Indifex and the Transifex Community
http://transifex.org/
http://www.indifex.com/
15 years, 1 month
Japanese UI Analysis, MacOS X Analysis, Action Items
by Warren Togami
Monday, March 16th, 2009 Warren Togami and Maureen Duffy conducted a
user interface analysis with a native Japanese user. Windows IME,
ibus-anthy and ibus-anthy were compared. The test subject is named
Kumiko, who formerly worked in a Japanese financial institution and is
familiar with ATOK on Windows XP. Our goal was to analyze the use
expectations of an "ordinary user" without technical background. Warren
Togami subsequently did some exploration of the IM software in the
latest version of MacOS X. Many bugs were filed from this analysis.
Further action items are proposed below but require discussion before we
can file bugs.
http://en.wikipedia.org/wiki/Language_input_keys
Describes Input keys on Japanese and Korean keyboards
https://bugzilla.redhat.com/showdependencytree.cgi?id=ibus&hide_resolved=1
http://tinyurl.com/fedora-ibus-bugs
ibus bug trackers
Japanese Specific Bugs
======================
https://bugzilla.redhat.com/show_bug.cgi?id=491042
ibus-anthy global activators by default: 半角/全角 or ALT-` or ALT-半角/
全角 or 無変換 + 半角/全角
https://bugzilla.redhat.com/show_bug.cgi?id=488666
ibus-anthy candidate selection should be vertical
https://bugzilla.redhat.com/show_bug.cgi?id=491040
ibus-anthy candidate selection lacks Pg-Up, Pg-Down and scrollbar
https://bugzilla.redhat.com/show_bug.cgi?id=490748
ibus-anthy katakana button doesn't do anything
https://bugzilla.redhat.com/show_bug.cgi?id=490747
ibus-anthy 無変換 Non-conversion button fails to undo conversion in
pre-edit buffer
https://bugzilla.redhat.com/show_bug.cgi?id=490750
ibus-anthy 変換 Conversion candidate button doesn't do anything
https://bugzilla.redhat.com/show_bug.cgi?id=491041
ibus-anthy Register Words interface missing
Korean Specific Bugs
====================
https://bugzilla.redhat.com/show_bug.cgi?id=487269
ibus-hangul is missing Hanja Conversion
https://bugzilla.redhat.com/show_bug.cgi?id=486056
ibus-hangul missing options for 2bul, 3bul and other Korean layouts
General ibus Bugs
=================
https://bugzilla.redhat.com/show_bug.cgi?id=490381
change "Next/Previous engine" labels to "Next/Previous input method"
https://bugzilla.redhat.com/show_bug.cgi?id=477918
ibus failures with pidgin?
https://bugzilla.redhat.com/show_bug.cgi?id=490013
ibus - Next Engine hotkey really good idea for anyone?
https://bugzilla.redhat.com/show_bug.cgi?id=490009
ibus - Deleting Next Engine shortcuts doesn't work
https://bugzilla.redhat.com/show_bug.cgi?id=491043
ibus Enable IM engine for particular language automatically
Other Bugs that need discussion
===============================
TODO: Implement standard template for help files for ibus, and localize
it (MacOS X has a great example of this done right)
TODO: CTRL-Space is annoying to Japanese and Korean IM, Propose
Inherited Hotkey Activator Plan
TODO: ibus-anthy Determine which options are important that need
configuration options
TODO: ibus-hangul Determine which options are important that need
configuration options
TODO: ibus-anthy Implement pre-conversion auto-learning predictive
candidate input (think cell phone, read below)
TODO: Investigate dead key fallbacks work properly?
TODO: ibus hotkey activators sometimes fail completely in gnome-terminal
or tomboy, while they work in other apps
Miscellaneous Notes
===================
• Kumiko did not knowingly change any settings, used default IM shipped
on her computer: ATOK.
• Windows IME was very similar to ATOK to the point she didn't realize
it was different at first.
* SCIM's interface was almost identical to Windows IME with button
behavior, with the exception of activator hotkeys being equally broken
from the perspective of ordinary user expectations.
Activator Hotkeys
=================
• Windows IME does not seem to have "Global" activator hotkeys. Hotkeys
change depending on which IM is active at a the moment in the language
bar. For example, if Japanese is the current active IM, then ALT-`
works as an activator but not CTRL-Space. Conversely, if Pinyin is the
current IME, CTRL-Space works. There might be merit in changing our
hotkey behavior to be like this instead of our global hotkeys that try
to please everyone, but end up annoying everyone. Discussion? (Note
below that MacOS X has no activator hotkeys.)
• ALT-` is bound to "bring up the text-mode menu" in emacs, but
otherwise not likely to conflict with anything.
• 半角/全角 button is exactly the same keycode as ` on US keyboard
layout. 半角/全角 however is a different keycode from ` on a Japanese
keyboard.
• 無変換 + 半角/全角 should also be an activator. The user held down 無
変換 and pressed 半角/全角, and was surprised that it did not activate
like it does on Windows. It is unclear if 無変換 is meant to be a
modifier like ALT, but in any case 半角/全角 should have worked in this
case while it did nothing with ibus.
• Simple Proposal: 半角/全角 or ALT-` activators by default because
doesn't seem to conflict with anything important. Meanwhile it is a lot
less likely to be activated by accident like CTRL-Space.
• Advanced Proposal: Hotkey Inheritance Model, to have particular global
hotkey defaults automatically only if you have those engines enabled
Candidate Selection Interface
=============================
• MacOS X Kotoeri
∘ Vertical with slider, hides less likely candidates as "More..." until
you arrow down and they unhide
∘ Multi-word conversion doesn't seem to support word selection with
left/right arrows like Windows IME or SCIM. e.g. ねこをかっています
then SPACE, try to choose between "have" and "buy", how? Kotoeri
Preferences -> Windows-like operation seems to enable the left/right
button to select which word to convert
• MacOS X Hangul 2bul
∘ Type character then Option-Return to pop-up Hanja candidate menu,
vertical with scrollbar like Kotoeri. To the right of each Hanja
character it displays in the standard Hanja dictionary format "Meaning
<Pronunciation>"
• MacOS X Simplified Chinese ITABC (seems to be Pinyin)
∘ Type character like "wo". SPACE shows you "Standard" candidates from
GB2312. Shift-SPACE shows you GB18030 candidates.
∘ Default is to display candidates horizontally, picking with
right/left arrow and ENTER, or number shotcut for quick picking. In
preferences you can switch it to vertical mode. In horizontal mode,
up/down goes to the next page of 9 candidates.
• ATOK Windows
∘ Kumiko said ATOK on her home Windows XP system also does
auto-learning/predictive input. Example: Type ま before SPACE
conversion and it displays words that you used beginning in ま. Very
convenient especially for typing unusual names repeatedly. She really
liked this feature of ATOK. This is similar to cell phone IM.
∘ Reportedly Microsoft IME lacks this predictive pre-conversion input.
∘ ATOK however lacks the helpful dictionary entries in the candidate
selection while displaying homophones.
∘ Proposal: Auto-learning pre-conversion predictive candidate selection
might be feasible to implement. We already have the anthy dictionary
and statistical algorithm which auto-learns. We seem to have the
building blocks to do this. IDEA! There is still time to propose this
as a Google Summer of Code project. A student could work on this as a
feature of anthy independently of everything else we need to fix.
MacOS X 10.5.6 (some interesting things we could learn from)
============================================================
• Each language Input Method has simple and succinct help pages
describing basic operation, localized into all languages. The help
pages are in a similar standardized format for each Input Method.
• Seems to have NO hotkeys for non-Native keyboards. Not even an option
to enable hotkeys. Eliminates possibility of hotkey conflicts that annoy
users. Mac users on US keyboards seem to rely on clicking on the
language menu in the systray to switch between English and Japanese
input. Individual IM like Kotoeri have non-default options to map the
caps-lock key to switch between Hiragana and Romaji mode without
switching away from Kotoeri. (Note: Some Japanese users on Windows I
have witnessed do not know that 半角/全角 is an activator hotkey, and
they similarly always click the language bar to enable Japanese IM.)
• Korean Types
∘ 2-Set Korean
∘ 3-Set Korean
∘ 390 Sebulshik
∘ GongjiCheong Romaja
∘ HNC Romaja
• Japanese Types
∘ Hotkey in menu
• Reverse conversion ^UP R (not sure what this means?)
• Retrieve fixed Yomi ^UP Y (undo conversion back to yomigana?)
∘ Hiragana
∘ Katakana
∘ Full-width Romaji (not default)
∘ Half-width Romaji (not default)
∘ Romaji
∘ Ainu (not default)
• Traditional Chinese Types
∘ Zhuyin
∘ Pinyin
∘ Cangjie
∘ Jianyi
∘ Dayi(Pro)
∘ Hanin (not default)
• Simplified Chinese Types
∘ ITABC (Pinyin)
∘ Wubi Xing
∘ Wubi Hua
Warren Togami
wtogami(a)redhat.com
15 years, 1 month
