At long last, I've completed the 0.6 release of the Tsukurimashou fonts (project home page). This one contains 1110 kanji, including all those taught in the Japanese school system through Grade Three. Also new in this release are experimental italics and integration with my IDSgrep structural-query software (which has its own, separate release series). Downloads: source code; precompiled fonts; demo PDF files.
At about six months in the making, this has been one of the hardest Tsukurimashou releases yet. I have some targets for what characters should be in each release; but for odd-numbered releases the target is generally "half of the characters on this grade-level list" and for even-numbered releases it is "all the remaining characters on this list that weren't covered by the preceding odd-numbered release." That means the characters in each grade that are difficult to code tend to get shoved into the even-numbered release, and the even-numbered release takes much longer to complete than the odd-numbered release.
I think my "even-release syndrome" is really the same issue that's responsible for many previous kanji meta-font projects being abandoned after about 120 characters. With 1110 kanji in version 0.6, I'm well past that limit and so I am pretty well sure that it is possible in principle to push Tsukurimashou to completion given enough availability of effort. But because I'm splitting it into small but rigidly-defined pieces (the school grade levels) I face the 120-character barrier in the middle of each piece. It seems like 120 characters (when you're choosing from a larger list) is about how many you can do before you start facing things like weird nonstandard relationships between character elements. Once you do hit that barrier, you need to have a Turing-complete definition language (which I have; many other projects didn't) and a fair bit more programming effort per kanji. Fortunately, there aren't that many grade levels left; the end is at least imaginable even if not in sight yet.
There are significant infrastructure updates in 0.6 as well. Most notably, Tsukurimashou now bundles the relevant code from METATYPE1, allowing it to compile on a basic Metapost installation without installing METATYPE1 and mtype13 as separate dependencies. There've been some fixes and tweaks to the build system as well.
I have some leads now on ways to make this work academically or even commercially relevant; it is possible I may actually have found a way to get paid to work on Tsukurimashou as a much larger fraction of my work time than previously. If true, that would be really neat. But not much is sure at this point.
I don't know if this project is really helping me learn to write kanji fluently or not, which was the original goal. It's certainly helping somewhat. But there are a lot of characters which I built into the font and then immediately forgot. I can't read let alone write all the kanji currently included in Tsukurimashou and I don't know that I could really match the average eight-year-old who has completed Grade Three. Then again, I have really high standards for when I will give myself credit for "knowing" a kanji. I am semi-seriously planning to take the JLPT level N4 this coming December and I guess at that time I'll find out how many kanji I really know.
As always, discussion and publicity would be very welcome. Post it, talk about it, like it, tweet it, or whatever, but please help me get the word out about this project.