Re: [edict-jmdict] kradfile expanded

[Ben Bullock <benkasminbullock@*********>]

On 09/06/2009, James Rose <jim@kanjicafe.com> wrote:

> I can't imagine a realistic scenario where the decision to revise the
> radicals listed after the "colon" in some random kanji could break anyone's
> (reasonably thought out) system.

Well, I don't want to say anything sarcastic, but instead of using
your imagination, you could try developing a system which uses the
data and then see what happens when the data is updated.

> The JIS
> x 0212 standard is effectively dead, and therefore not going to change, so
> the kanji inside of kradfile2 will not change - so I think the fear of
> revision should not exist.  This is not complicated enough data to merit a
> version control regime in my opinion.

Another advantage of a version control system is allowing people to
update the data themselves. When I was developing my multiradical
system I found lots and lots of minor errors in the radkfile data (all
reported to Jim Breen). Since these were just found from random
clicking, I'm sure that a lot more errors remain. Wouldn't it be
easier to have a way to send patches which could then be accepted or
rejected? I know this kind of thing is possible with github at least -
you can see lots of people doing minor fixes on other people's work as
they find them, which is very nice since it saves the communication
overhead of sending an email plus reply for each little error.