I spent a little while yesterday poking around with getting Japanese input to work on my home Linux machine, since I’ll need that for entering vocabulary cards into the memory program.

To make a long story short: largely, it Just Works. (At least under Ubuntu 8.04.) I was a bit confused at first by the wealth of possibilities: in particular this page talks about UIM (better but less standard?) versus SCIM. I compared the pages that it pointed at (both for 7.10), and then I found this page which was an updated version of the SCIM recommendation for 8.04; reading through that, it seemed like some of the UIM advantages had gone away, so I decided to go with SCIM and follow the steps listed there.

I made it as far as turning on Japanese language support (including the support for entering complex characters), and got confused at the next line on the HOWTO: it was directing me to edit a file in my home directory which didn’t currently exist (so why not ask me to create it instead?), and it was telling me to edit it via sudo, which didn’t make any sense. While I was googling to make sure I wasn’t about to do anything stupid, I ran across this page on SCIM, which claimed that things should work now with no further configuration.

So I tried typing Control-space, and up popped a SCIM window; a bit of typing later, and I had both kana and kanji appearing! Pretty cool. And the input method does seem quite usable; I didn’t have to do to much experimentation before I felt like I could enter characters fairly reliably, without too much extra typing. One thing which I might want to tweak is using a key combo other than Control-space to enter it, because I use that a lot in Emacs; fortunately, SCIM only recognizes the actual control key instead of the caps lock key that I have remapped to control (perhaps related to this bug), so the default mapping isn’t interfering with Emacs use; if it starts doing so, it should be easy enough to change.

The upshot is: under Ubuntu 8.04, Japanese input Just Works, as long as you have it turned on! I wish I’d found that latter page first, rather than the longer page. For now, I’m not doing anything more with the longer page (and am considering removing the extra repository it had me add); the next thing on its list is a handwriting recognition system, which sounds cool but which I have no particular desire to use.

I am considering following its font instructions; in particular, I agree that it’s annoying that kana look blurry at small sizes. Having said that, for now I’m holding off: my main planned use for Japanese input is in the memory project, and I’m planning to use larger font sizes for the flash cards there, so it may not be a big deal. And these days, I’m generally in the mood to go with defaults whenever possible: I have other ways that I’d rather spend my time than fiddling with system configuration.

One thing which it pointed out which is useful: go to System / Preferences / Appearance / Fonts and turn on subpixel smoothing. (It also recommended clicking on Details and making sure hinting is set to ‘full’, which was already the case on my system.) That noticeably improved the look of my (roman character) fonts on my monitor; if you’re using an LCD monitor, you might want to check that your system is configured that way, even if you don’t care about Japanese at all.

Edit: I’m no longer so convinced of that last sentence: fonts started looking a bit green when I did that. It’s worth trying, but I’m thinking of going back to ‘best shape’ instead.

Post Revisions:

There are no revisions for this post.