This may duplicate some of the above, I hope a record of my observations will help. On Firefox 3.0.10, and MAX OS X 10.5.7 PDF links get opened in an Acrobat Window, and not in the browser (I think this is normal Mac Firefox behaviour), perhaps i need a plugin. Any pdf url dropped from the browser into Zotero, opens in acrobat, and doesn't leave an item in Zotero. The resultant downloaded file does not have an acrobat icon in Finder, but opens Acrobat when double-clicked. If I drop any (with or without icon) pdf into Zotero from Finder, a non pdf item gets created, metadata retrieval menu not available.

Thanks Dan. Keep up the good work...

which issue exactly are you referring to? The one mentioned by pjgoodall (which should be fixed) or the one noted at the top by Vincent (which is current intended FF behavior)

if you are using pdfs from journal articles - why don't you use regular translators?
I.e. the icon in the URL bar, e.g. for JSTOR, ScienceDirect, Informaworld etc.

Also, for most pdfs you download from the internet, right-clicking and selecting "Retrieve Metadata" will not just create a parent item, but also include the relevant citation information. I think this was the main reason to change this in Zotero:
Otherwise, to use the "retrieve" function, you would have to create the item, drag the pdf out of the parent, delete the parent, then select "retrieve metadata" - which, in turn, many people considered quite annoying.
I haven't seen any reference to this being changed in the near future - the behavior for "retrieve metadata" I believe will be adjusted so that is produces empty parents when it can't retrieve the information.

Yeah, clicking the browser icon is a handy feature when it's available and I do when I am using Jstor and the like. Unfortunately, this particular bit of research I'm doing has me searching for 2006-2009 articles and Jstor only covers through 2005. For articles that recent I pretty much have to go to the journal websites directly and icons are rare. Further, "retrieving metadata" results in getting stuck with no data and no way to cancel the process. I have to restart Firefox to get out of it. Big Pain!!

Still, I wanted to test adding an item from Jstor to see if I had the same problem and there are some definite issues.

Problem 1: The icon is available in the Jstor page viewer. Click it, it stores metadata just fine, but no link to the original article, no pdf, etc.

Problem 2: Next, I tried opening the pdf from Jstor and there's no icon in the browser then. So I tried saving the item from current page and like always it creates a standalone pdf. Okay, retrieve metadata, and it does what it should, creating a parent with pdf attached. However, it was the wrong metadata!! Instead of "Reinventing WAC (Again): The First-Year Seminar and Academic Literacy" which was the article I was actually viewing, it retrieved info for something called "Raised levels of a new immunoglobulin class (IgND) in asthma" and recorded jibberish as multiple authors - looks like it might be the tags or keywords, but definitely not names of people responsible for writing the article. That's not going to work either.

So, I say the saving a pdf process still needs a bit of work. Any idea when that might happen? I love Zotero, but I can't afford to let it slow me down at this stage in my research. I would appreciate any help.

Problem 1 shouldn't exist. If you have institutional access to JSTOR and have the "automatically download pdf" option checked in the preferences you should get the pdf.
If not, post a link.

Problem "0" shouldn't exist either - clicking on cancel should stop the retrieve metadata process without restarting firefox. Please provide steps to reproduce that error.

As for Journal sites - which one are you referring to? - Zotero has translators for all major Journal Publishers including Springer, Elsevier, Muse, Cambridge U Press, Taylor and Francis, Blackwell, in addition to that Ingenta, EBSCO, Proquest, Sciencedirect and several others. I pretty much never encounter a published journal article that I cannot import using a Zotero translator.

The last problem is known and improving that functionality is relatively high on the agenda I think - the most frequently discussed option is a) to include what Bionatsci says above and b) check with the user before actually importing the metadata to see if its for the right article.