F.5.3 Outstanding issues

This is the to-do list. These are the issues which should either be addressed before the next full release, or postponed as being unresolvable just now. For more wishful thinking, see Section 8.1.

Bundle Jade with the system

At present, the distribution consists only of the SGML library stuff. I need to work out the magic to get Jade to build on Suns and Alphas, and then distribute them together. Since Jade is needed to build the documetation during installation, you can't even install the package on systems which don't have Jade. I therefore haven't tested the installation makefile other than on Linux: Tru64's make is dumb enough that I can just about guarantee that there'll be problems there.

Finish documentation

Read-through for style, plus more expansive installation notes.

Use of FPI: should we use -//Starlink//... or +//IDN star.rl.ac.uk...?

The second is a standard `registered prefix' built up from a DNS domain name. Is there any prospect of a starlink.ac.uk domain name?

Include some facility for dealing with obsolete exported IDs.

One possibility (discussed at the programmers' meeting) was <sect id=newid>Intro<oldid id=oldid>, but that's not very flexible - it'd only apply to sections. I thought instead of

<!element oldidlist o o (oldid+)>
<!element oldid     - o EMPTY>
<!attlist oldid
	id	id	#required	-- the now-obsolete id --
	newid	idref	#requred	-- replacement --
	-- resolve a reference to this element as if it were a
           reference to the element pointed to by newid, but
	   warn or issue an error about a deprecation --
	export	(export) #fixed export	-- for compat/consistency? --
	>

This means that at the top of the document (probably), you could include a list of elements:

<oldidlist>
<oldid id=old.id newid=replacement.id>
<oldid id=another.id newid=another.replacement>
</oldidlist>

This is both perfectly generic, and usefully visible.

Bugs in Jade's LaTeX back-end

There's a deficiency in the LaTeX back-end I added to Jade. There's some weakness in the way that characters are output, which results in SDATA entities not being processed properly, when the character data is obtained via (data) rather than (process-children). Unfortunately, I think that fixing this one is going to need more knowledge of Jade internals than I have to hand right now. The only available workaround is to avoid using any of the entities in Table 1 within `DT' elements (which is the only place the stylesheet uses the feature which triggers the bug). If you fall victim to the bug, the only effect would be a &#65533 or \\char65533 in the output, which might confuse a browser or LaTeX, but do nothing worse.

Bugs in the mk script

There are currently a few missing targets in the installation makefile.