Harvard style testing

dstillman

Codec created some Harvard style CSLs. Anyone who wants to test them can follow these steps:

1) Go to the ticket.

2) Click on one of the two attachments, harvard1-raw.csl or harvard2-raw.csl. Scroll to the bottom of the attachment page and click "Original Format" under "Download in other formats".

3) Open the downloaded CSL file in a text editor (e.g. drag into Notepad or TextEdit) and copy the contents to the clipboard.

4) In a Zotero-enabled Firefox, load chrome://zotero/content/tools/csledit.xul. Open the Zotero pane and select some items. Paste the CSL contents into the large text area, and you should see a preview of the items in the Harvard style.

Jon Rubin

Hi Codec
Are the files the same?
They both have the same id
<id>http://www.zotero.org/namespaces/CSL/Harvard2-Author-Date.csl</id>;
<source>http://www.leeds.ac.uk/library/training/referencing/harvard.htm</source>;

are they the same Leeds and Anglia? I haven't checked all the way through but I assume there's been an uploading snafu.

One thing: on the bibliography, the et.al. rule is set to one author. The Leeds website isn't exactly helpful here (is it implying that the et.al. should hit for au>1, in which case, apologies your rule is right - as I said I don't think Leeds is clear), but the Anglia one says that it should be 4, which seems about right.
If Anglia and Leeds are different, I'm happy to test the Anglia one too. Really grateful for your work here, Harvard is such a useful stye - even if it isn't terribly standard!

A slight side point/question - does any Harvard style that you know of reference the translator of a work ever? The Leeds and Anglia don't seem to.

Regards, Jon.

Codec

Oops - yes a snafu in production - fixed now.

I deduced the et-al rule on Leeds, I think Anglia is set differently to 4 as you say.

Thanks for the feedback.

Jon Rubin

HI Codec, thanks again. I'm going to concentrate on the Anglia - just because I find their documentation easier to follow.
I tried putting two books by the same author, different years. Anglia says this should be
(Bloggs 1992, 1994)

Yours has it as
(Fulford 1982; Fulford 1989)

For journal articles and book chapters it should be "p" not "pp"

For book chapters of edited collections there should be a full stop after the chapter title and so the "in" should be "In" and there shouldn't be a comma after ed.


Regards, Jon.

bdarcus

Is there not one definitive reference for Harvard? Googling suggests perhaps not, but it would be nice if we could clarify ideally one "Harvard" style.

Also, on this process ...

First, manually downloading and copying-and-pasting files is too much hassle for my busy life. It would be easier if you loaded styles as individual files (rather than embedded in a SQL file) in a repository I have both read and write access to, so that I can more easily help improve the style(s), and then just link to there. Having to deal with patches is a PITA.

Obviously the xbib repo has these characteristics, and I've been happy to give people write access to it (though I wonder if a more distributed SCM like mercurial is more appropriate for style development).

Second, I've asked a twice recently for people to state somewhere what styles they're working on, but I've not heard anything from anyone but codec. Am I really to assume the Zotero team isn't working on any styles?

For the record, I have started on Bluebook, and have made incremental progress.

bdarcus

BTW, I just validate one of your files (1); it's not valid ;-)

Codec

hehe
I just validated them, it found two errors.
One was the sort-algorithm wasn't a legal thing
The second was the class="author-date" which it seemed to object to.
I used the emacs nxml mode to verify

What did you find?

bdarcus

The same. (nxml mode is excellent!)

Codec

Thanks Jon for the feedback - just uploaded a new version which I think fixes these problems (and hopefully doesn't introduce any new ones!) and IS validated. :-)

Jon Rubin

(just looking at the Anglia version)
Its still showing "pp." for journal articles, it should be "p."
The other stuff seems fine :-)

Codec

Fixed - thanks again!

Jon Rubin

Great stuff - when can it be rolled out for us techno-peasants that don't like / know how to use SQiite to manually import it?

dstillman

when can it be rolled out for us techno-peasants that don't like / know how to use SQiite to manually import it?

As soon as you guys say it's ready.

bdarcus

And can someone please address my questions/concerns above?

Among other things, would it be possible for you to give these styles cool URIs; e.g. something like and to actually make them get-able (even if Zotero doesn't yet load them from URI)?

dstillman

Bruce: We know the process is lacking. I'll try to rig up something this weekend to help. More later.

Jon Rubin

Hi Codec, have you tried inserting the csl into your Zotero SQLite database? I only ask because I tried and although I can see the style to select, I get a commincation with Zotero error / can't drag a citation with this style selected (but can with others).
In all liklihood I haven't inserted it right and if you've got it working then its definitely me.
FYI found an opensource SQlite browser-GUI which made inserting the new record appear easy - much easier than mucking about with SQL - if it weren't for it not working that is :-)
http://sqlitebrowser.sourceforge.net/development.html for the GUI
Regards, Jon.

Codec

Now you mention it - yes it does give word a problem.
I think I've found the problem and fixed it.

I wrote a small dos .bat script to install the new formats automatically, but its rather fragile due to the firefox profile directory being a little elusive sometimes. It only worked too if you had the default location for the database.

Jon Rubin

Well, I'm using openoffice, so if we can get it working it'll definitely be ready to release.

Still can't get your script to work (still have the can't communicate with Zotero problem).
So I tried copying the csl field from the chicago author-date record and it worked fine. So I guess there is still a problem with the script but on the plus side (for me) the SQgui works fine. Sorry I can't help more.

Codec

What csl line did you use?

Jon Rubin

Well, its working now! I think I may have had the wrong cslID before (so it was me, no surprise there!). Now I've got:

http://www.zotero.org/namespaces/CSL/Harvard1-Author-Date.csl,2007-10-23 18:00:00,Harvard1-Author-Date.csl,<?xml version="1.0" encoding="UTF-8"?> ...

Great stuff.
Two things though:
pp. for edited collections and journal articles, should be p. It's right in the bibliography but not for the in-page citations.

and I just noticed and I'm not sure if CSL can handle this or not. When citing a page range that's more than 3 digits it should be e.g. 1557-70 not 1557-1570. Of course I can change this in the Zotero database but then would have to change it back if using a different style.

Codec

Great - I'll fix the page number thing.

bdarcus

and I just noticed and I'm not sure if CSL can handle this or not. When citing a page range that's more than 3 digits it should be e.g. 1557-70 not 1557-1570.

No, CSL does not handle this, because it's a large can-of-worms. These kinds of shortening algorithms are often quite complex (read the Chicago guide, for example).

I think, though, it might make sense to have this an option, where there are different name rules. I'd be happy discuss adding this with the other people involved in CSL, but I'd like if someone could do some research and find out if there are good values to add other than "chicago."

Jon Rubin

Hi bdarcus, I'm not surprised CSL doesn't handle this for the reason you mention. I remember a comment you made ages ago about abbreviation (including things like ibid and op cit) being a legacy of the pre-electronic era. Personally I'd be glad to see it go (along with endnotes - don't keep making me flip to the end of something to see what you're referencing).
Also, to make things harder the alternative (Leeds) Harvard that Codec has set up doesn't seem to shorten page numbers in this way but does use pp.
WALKER, J R. 1998. Citing serials: online serial publications and citation systems. Serials-Librarian, 33 (3/4), pp.343-356.

Jon Rubin

Hi Codec, hopefully one last little thing. When you supress the author, it inserts a space between the first bracket and the date. I noticed it does this with the APSA format too (I was flushed with the success of getting the Harvard1 citation working, so loaded that one too but haven't really had a chance to examine it closely).

Regards, Jon.

Codec

Thanks - think I've now fixed those too.

zinaf

I have two sources, from my uni (open.ac.uk) and another from the web, that both state that in the reference list you use full page numbers and pp (e.g. pp.343-356).
I think there are some dicrepencies out there. funny enough i never managed to find info from harvard themsleves :)
I vote for keeping the pp and full page numbers for the intervals (p still used for referring to a single page though)
I will try now to test the citation format.
cheers

zinaf

I have done the testing and it works. I have however comments on the way the citation is organised.
In-text citation - author and date should be separated with comma.
Reference list - for journals - journal volume and issue number should be structured differently according to the two sources I have have, i.e. 'Vol. 1, No. 3' and not 1(3).

I think, the only viable solution to the problem of not having one Harvard standard is through the new feature in zotero that would allow us to edit easily the citation formats to make it match the requirements of the institution we are affiliated with.
I reckon you could roll out this citation the way it is since there will be no agreement. I should be still usefull until we get the new feature in zotero.
cheers. thanks for the efforts. much appreciated.

Codec

I've done a custom format for an open university course I'm on (A251) although most of mine are S courses. This course has its own weird rules, but has some Harvard like syntax. Some of the rules are specific to this course, so I don't think its worth releasing it as there is a substantial amount of hackery required to get it to work (abusing certain fields to get citations working correctly).

I suspect an easy to use GUI is not going to be very easy to make. There are quite a number of decisions to make in layout and even then its not always possible to find rigorous rules. However if we can get plenty of examples done, and an easy way to add new ones, I think eventually pretty much everything will be catered for, or else close enough that a couple of small tweaks are all that is needed.

Jon Rubin

HI Codec, still getting the space between the first bracket and the date when I use supress author.

Jon Rubin

Hi Zinaf, there are definitely different Harvard standards out there that is why the two that Codec has created both have urls to let their users know where to check the standards against. If you'd noticed my comment earlier I pointed out that the Leeds standard does use pp and does not shorten the second page range. I haven't tested it yet but I'm guessing it will provide some/most of what you want. There isn't (as far as I understand it) any problem with having diferent Harvard styles available, so long as they're clearly identifited. I'd suggest that Harvard1, Harvard2 etc. isn't the best long term solution :-) Maybe Harvard+institution? So at the moment you'd have HarvardEA, HarvardLeeds, HarvardOU etc.

I had a quick look at the OU style guide http://library.open.ac.uk/documents/Harvard_citation_hlp.doc
and there is a wonderful discrepancy to keep us all on our toes between print journals which cite vol. 8, no. 4 and ejournals which cite 8(4).

bdarcus

I've done a custom format for an open university course I'm on (A251) although most of mine are S courses. This course has its own weird rules, but has some Harvard like syntax. Some of the rules are specific to this course, so I don't think its worth releasing it as there is a substantial amount of hackery required to get it to work (abusing certain fields to get citations working correctly).

Hmm ... "hackery"? Does this suggest you might have some suggestions for schema tweaks?

On releasing styles: I've mentioned to Dan in a private email that where I want to get (sooner rather than later) is you create a style for some course or journal, post it at some stable location, and your colleagues can simply load/subscribe to that style by clicking or pasting the style URI into Zotero (or any other application that supports CSL). A user should never -- ever -- have to manually distribute or load a CSL style. Atom will make this process smoother, allowing browsing of collections of styles and such, automatic updating, etc..

Right now we're in a strange initial stage where styles (in Zotero) are all or nothing: either everyone gets access to them, or nobody does. But I'd encourage you to make them available, even if they don't get loaded in Zotero by default. "Available" means retrievable over HTTP either by the value of the cs:id element, or the cs:link.

And yes, a full GUI will be difficult; the power of the markup gets a little hard to represent in a GUI. But, one can think about something like BibTeX's makebst: a wizard that allows you to create an almost-perfect finished style based on a series of questions.