Again on the topic of URLs in bibliographic info for print -based materials:

(1) They are nice to have in your metatda, since you can use Zotero's view button to get quickly to whatever the URL points to: a database record (possibly with fulltext), an online version of the reference, whatever.

(2) I don't want them in my citations, since most of what I'm doing is all old-school pages-and-binding citation.

(3) I stick to my original impulse above. Could there somehow be an option to keep URLs *out* of citations of traditional print-based materials. How is it best to do this? (Create 2 versions of every CSL stylesheet? Include a checkbox in Zotero's preferences?)

(4) At the risk of over-generalizing from my own case, I would suggest that for most non-web applications, citing print-based resources without the URL is still the desired behavior.

Very good. Thanks.

"The relevant URL should be included by default."

Yes, but only if that is the version consulted. I'm mainly talking about two use cases:

1. The metadata is from an online database like EBSCO, and the URL for the database entry is imported into Zotero, but I consult the paper edition. In this case I have the URL but it should not be included.

2. I have both the metadata and PDF from an online database (like JSTOR). The url to the JSTOR main page for an item (not the PDF itself) in imported. I consult the PDF. In this case perhaps the URL should technially be included but I don't want to do it. And I don't think it's necessary since the PDFs are nearly always scans of the hardcopy pages, or in the case of newer issues, they seem to be the camera ready copy produced for the printers (though they may in fact be proofs). I still want to use the 'old fashioned' way and cite them as if I had used the print edition, since I consider the differences to be negligible.

Scot: The issue with how URLs are handled is a bug with Zotero, not with the CSL style. It's important to keep in mind that types like book, article and chapter serve as generic fallbacks. So they need to include the URL by default. What Zotero needs to fix is how the translators work such that they by default add links for items in journal databases, rather than populate the URL field.

Bruce: ah, right. That's clear enough. Links to database records shouldn't go in the URL field at all. The field is actually for 'the URL to this item.' Does this mean that a link to a JSTOR pdf (which is, say, a scanned copy of a paper article) should also not go there, since when I cite that entry I mean (very traditionally) to cite the paper version (of which the PDF is a copy). This sounds like an obscure case, but it's actually what I do fairly frequently. I suppose in that case the PDF should only be added as a child link.

Do I have this right then that (according to the CSL spec) no paper-and-binding items should have anything in the URL field at all, except in the case that (1) they also exist in a web-reachable format and (2) that is the version I want to cite? This of course makes Indi right.

It is a shame that, as erzalogo says, adding child links slows things down. My database is only 1000 top-level items, but already it's getting painful on my 900MHz thinkpad laptop. (It expands to 1300 items) with children. I'll have to loose some of those less essential child elements eventually, I suppose.

It sounds like the solution is:

1. Modifiy the translators and scrapers so that they don't put anything but a URL to the item itself into the URL field. (no links to database entries).

2. Add a pref to turn off adding child links for items added from databases (to keep database size down)

3. Add erazlogo's user defined URL field on the Info pane.

4. Possibly add a pref to turn off URLs in citations, as we discussed above, though it seems like if the CSL spec is followed, we won't need that as badly. (or, Bruce, what do you say?)

5. Ideally, help us get the links to database entries out of our URL fields. (Global search and replace?, Global field editing?)

Comments?

The developers have said they intend to sort this out better, which I'm sure will happen. It's connected to a set of larger issues about how to keep URIs, and what should count as canonical URIs for an item, and how to get your citation style to fall back to intelligent ways of 'pointing' to an item, all while staying within (or nearly within) existing print-based citation styles.

The solution in the short term is to keep those URLs out of the item's URL field (by manually erasing them, or, if you want to keep them, by putting them in a note). I usually do that as I quickly check the item's metadata just before citing it. That way I only have to do it to the things I actually use. JSTOR's metadata isn't of bad quality, but I've imported heaps from lots of sites, and I rarely have an item where I don't have to fix some little thing. This is just another one, for now.

Go to Preferences > Export and uncheck "Include URLs of paper articles in references" - that will take care of JStor articles.

As we used to say, "Jinx erazlogo. You owe me a coke."

Thanks for pointing out that option... my problem seems to be that the zotero plugin for word (which I am using here and all comment relates to the citation and bibliographic data inserted by this plugin) isn't honoring it. I'm sure I am up-to-date with the latest official release.

More...
1. I already had the "include URL" option unchecked in my preferences panel
2. checking and then unchecking does not alter the behaviour
3. The URL appears in both the word bibliography (generated by the plugin) and the bibliographic entry generated from the right-click menu in the main zotero interface (in firefox)
4. The behaviour is not consistent. All entries in my database have a URL field however it is included in the bibliography only for some (unfortunately most).

Looking more closely there may be some deliberate effort here to use the URL in cases where a publisher (or sufficient identifying information) is thought to be missing. Perhaps my problem comes back to bibtex import.

Will keep posted. Thanks... sorry about the coke ;-)

Thanks for pointing that out! I feel blind now :-(

This is definitely what is going on then as most of my bibtex entries do not contain page numbers. Unfortunately the also often include URL fields which are only relevant on my intranet. Not a typical zotero workflow. I'll have to find a fix somehow.

On a general note it makes sense to me to have the bibliography style determine whether or not a URL is included and perhaps to have this as an option to the style.

Can I somehow hack the bibliography style definitions?

It shouldn't do that and I cannot replicate this. Which Chicago style?
how does it look in the test panel?
chrome://zotero/content/tools/csledit.xul
What item type?

Sorry--just pasted the URL. Full test pane output is
Single Citations
James Marriott, Plan of a Code of Laws for the Province of Quebec (London, 1774), http://find.galegroup.com/ecco/retrieve.do?contentSet=ECCOArticles&sort=Author&tabID=T001&searchId=R1&retrieveFormat=MULTIPAGE_DOCUMENT&prodId=ECCO&currentPosition=1&userGroupName=29002&resultListType=RESULT_LIST&sgHitCountType=None&qrySerId=Locale(en,,):FQE=(TI,None,84)Plan+of+a+code+of+laws+for+the+Province+of+Quebec;+reported+by+the+Advocate-General.:And:LQE=(BA,None,124)2NEF+Or+0LRH+Or+2NEK+Or+0LRL+Or+2NEI+Or+0LRI+Or+2NEJ+Or+0LRK+Or+2NEG+Or+0LRF+Or+2NEH+Or+0LRJ+Or+2NEM+Or+0LRN+Or+2NEL+Or+0LRM$&inPS=true&searchType=AdvancedSearchForm&&docId=CW3324042934&retrieveFormat=MULTIPAGE_DOCUMENT&docLevel=FASCIMILE&workId=CW3324042934&relevancePageBatch=CW124042934&showLOI=&contentSet=&callistoContentSet=ECLL&docPage=article&hilite=y.

Thank you for clarifying -- much appreciated.