References to these must show title, institution, date and issue number (edition, if you like). I can't see a document type that gives me these fields. Or is it possible to create new document types?

Yes, as far as I can see it lacks an issue number (or edition) field.

Yes, but the report number is presumably a reference number, not an issue number. Anyway, I need both, and it can't be both.

Look for example at the document "ESARR 3 Guidance to ATM Safety Regulators : Explanatory Material on ESARR 3 Requirements" at http://www.eurocontrol.int/src/public/standard_page/esarr3.html. The document number is "EAM 3/GUI 1" and the issue number (shown in the document as "edition") is "1.0". The issue number isn't necessarily purely numeric, by the way .

Another problem with the "report" document type is that zotero doesn't seem to have any mapping for the "institution" field (at least, http://dev.zotero.org/csl_syntax_summary doesn't show one), so I can't include it in a reference. In the example above I could kludge it by putting the institution in the "author" field, but if I need to reference a technical paper by author /and/ institution -- common in engineering -- then I'd be out of luck.

"ESARR 3 Guidance to ATM Safety Regulators. EUROCONTROL, Document EAM 3/GUI 1, Issue 1.0, 1st June 2001"

I think the issue number field would have to include the whole text "Issue 1.0", because I might have to refer to a draft, such as "Draft 1.2".

I call it "version", but I'm not the entire community :-) EUROCONTROL uses "edition", probably because it's closer to the French...

Identify the style you need modified.

I finally worked out how to hack the csl files... Here is a patch for the ieee style that reports the details of technical reports (and some other small mods). The base style is fairly simplistic so I can't guarantee there will not be cases where it misbehaves... but for me it seems good. This patch includes in the bibliography (sorry whitespace seems to be removed in post):

1. report type and number
2. month (short form) as well as year for book and chapter types
3. only includes the URL if the type is webpage

--- ieee-modified.csl Fri May 16 13:32:11 2008
***************
*** 47,53 ****
--- 47,56 ----
<macro name="publisher">
<text variable="publisher-place" suffix=": " prefix=" "/>
<text variable="publisher" suffix=", "/>
+ <text variable="genre" prefix="" form="short"/>
+ <text variable="number" prefix=" " suffix=", " form="short"/>
<date variable="issued">
+ <date-part name="month" suffix=". " form="short"/>
<date-part name="year"/>
</date>
</macro>
***************
*** 105,111 ****
</group>
</else>
</choose>
! <text macro="access" prefix="; "/>
</layout>
</bibliography>
</style>
--- 108,118 ----
</group>
</else>
</choose>
! <choose>
! <if type="webpage">
! <text macro="access" prefix="; "/>
! </if>
! </choose>
</layout>
</bibliography>
</style>

Looks good, except for the 'webpage' conditional. If I cite a report or an article that's published at a URL, are you saying that IEEE does not want that URL?

Also, what happens if you remove that 'entry-spacing' option altogether?

BTW the IEEE style manual does explicitly talk about online sources from page 10.

"C. On-Line Sources
The guidelines for citing electronic information as offered here are in modified illustration of the adaptation by the International Standards Organization (ISO) documentation system and the American Psychological Association style."

It might be possible to do something meaningful using substitutions in the CSL. Not for me at this time though I'm afraid.

for more discussion on the extra space see here http://forums.zotero.org/discussion/2252/

WRT every document being a web document I respectfully disagree... let me explain my perspective:

Just because a document is made available on the net doesn't make it a web document. What is far more important is who the publishers are as that indicates the review process, context (conference vs journal etc) and the research community in which it circulated first. If you google a title you will often find a handful of web locations for the same document, but only the publishers web address would be deemed appropriate for inclusion in a citation (unless you are specifically referring to a pre-print). As a result this is adding no new information. It is just like saying "Available in the printed transactions at your library". (Wo... starting to rant there... please excuse)

If there is no "formal" publisher then a web address (or in the bibtex world a "howpublished") becomes important. Alternatively, newspapers and such often maintain separate web versions that may be slightly different, making the distinction between printed and web versions important.

I am feeling old-fashioned making this case but I feel this distinction is real and important (and I'm a pedant). IMHO citations should not be verbose for the sake of it but neither should they be lazy. Universally employing a URL facilitates both these errors as
- if you include complete readable publisher info the URL is redundant
- if you can't be bothered collecting all the publisher info the URL is in some sense sufficient as you can find out the other information

Hope we can find some points of agreement. I'm so enjoying being able to finally bring some of the simplicity of latex/bibtex to word (in which I am forced to work).

"The purpose of citation information, after all, is to make it easy to find things, and there's nothing easier nor more direct than a link directly to the source."

The purpose of citation IMHO is threefold,
1 to allow a future researcher to find the item
2 to acknowledge the work of the author or source
3 to provide legitimacy (via academic gate keeping - peer review, institutional legitimacy - etc) for any use you make of that author's work to provide verification for your own argumentative claims
There may be more but I can't think of them.
My point however is that all of these create slightly different but overlapping requirements of a citation instrument.
The argument above appears to stem from a difference of opinion over the relative emphasis to put on point 1 or point 3
JT