[erlang-questions] Java error making docs in R15B01

[Richard O'Keefe]

On 7/06/2012, at 11:59 PM, Michael Turner wrote:

> "By the way, are the DTDs used in building Erlang/OTP documentation
> *documented* anywhere?"
> 
> Didn't Ulf just point them out on another thread?

E-mail is a *distributed* system.  Ulf's message saying where to
find the documentation arrived in my mailbox the day after I sent
my question.

The right thing, of course, is for each of the DTDs to contain
a comment saying where to find the documentation, or better still,
for the documentation to be automatically extracted from comments
in the DTDs.

There are some issues that are not documented.

When I've set up DTDs and had a use for something that was also
found in HTML, I have generally use the HTML names for things.
Thus when I have a table, I use <table>, <tr>, <th>, <td>, and
so on.  When I'm told to use <row> instead of <tr>, I want to
know why.  Perhaps some other template is being copied?
DocBook maybe?  It would be useful to know, because I might find
useful background in the thing that was copied.

Similarly, the HTML 3.2 character entities are copied, well and
good, but why not the HTML 4 ones?  I'm thinking of section 24.3
"Character entity references for symbols, mathematical symbols,
and Greek letters" with pubid "-//W3C//ENTITIES Symbols//EN//HTML".
If we're using xsltproc and FOP, which use Unicode, what's wrong
with these Unicode characters?  Can I use numeric references to
get them?  If there's some basic limit here, if, for example, it
is still thought important to generate pure HTML 3.2, I'd like
to know.  Or is it a limitation of UNIX man pages?  Need that
limit apply to documents that won't be processed by troff?