Re: [edict-jmdict] Changing entities to attribute values

[Glenn Maynard <glenn@********>]

Oops--there's already an element named <pri>. <freq>?

Here's some minor notes--may as well bring them all up together:

* <xref> and <ant>'s "center dot" notation is a little odd for XML;
splitting this apart avoids the need for parsing.  <xref keb="僕"
reb="ぼく" sense="1"/>

... though it'd be nicer to reference by ID, since that would simplify
finding the entry.  It'd also make reverse lookups easier (finding all
entries that reference an entry).
<xref id="1521400" reb="ぼく" sense="1"/>  Maybe reb is unneeded this
way; I don't know all of the uses.

* <bibl>: <bibl type="wikipedia" id="Japanese_language"/>.
* <links>: <link tag="image" url="...">Picture of a fish</link>, or
leave the description in a sub-element if you want to allow
translations for that.
* Audit: <audit date="2010-06-03">Entry amended</audit>
* The re_nokanji element can have #PCDATA removed, and then there's no
need to say anything about "usually have a null value".
* Looking more fundamentally, I'd probably make reb and keb
attributes, since they're effectively the key of each k_ele/r_ele.
<k_ele reb="かんじ">

Related schema question: are k_ele and r_ele guaranteed to be unique
within each entry?  I don't think there's any way to specify that in a
DTD, but it's useful to know (and I'd have to contrive pretty hard to
think of a reason they wouldn't be).

-- 
Glenn Maynard