Translator for PubMed intermittently fails

james.k.russell

example: http://www.ncbi.nlm.nih.gov/pubmed/22322284

Access to Amazon for a book, e.g., working. Access to PubMed (which I use frequently) often comes up with no citation download icon available. Rebooting my PC typically fixes this, but if I have to reboot just to download a citation, the service is pretty much practically worthless to me. Would be a big shame - I love Zotero when it works.

dstillman

We'd need a Debug ID for a page load for which the icon doesn't appear, and more info on what you're using (i.e., Zotero for Firefox or Standalone + a connector).

adamsmith

Since this may well be indicative of a pretty major problem - I'm not aware of any other report of this - definitely what Dan says, we want to debug this.
But just FYI if you're in a hurry, you can also just paste the PMID (i.e. 22322284 in your example) into the add by identifier (magic wand) function in Zotero to get exactly the same import as from the database. It takes a little longer, but certainly better than a reboot.

james.k.russell

Dan - I use Zotero Standalone with a connector from Firefox.
I tried to help with a Debug ID, but as there is no download citation icon in Firefox in this condition, there is no way to trigger debugging - neither Zotero nor Firefox "knows" that I would like to download the citation, and I don't know of anyway to tell them.

Adam - thanks for the tip - I wasn't familiar with the magic wand function, it works nicely for this purpose, though a bit laborious when I want to enter a bunch of findings from the PubMed clipboard, for instance.

adamsmith

re-read what Dan asks you to provide. We need a debug ID that covers loading (or re-loading) the page in Firefox - that's when Zotero should recognize it and show the URL bar icon, so we need to look at the debug during for that process. (edit: and we need the debug from Chrome, not from Standalone in this case).
I'd also like version numbers for Firefox, Zotero for Firefox, and Zotero Standalone

dstillman

as there is no download citation icon in Firefox in this condition, there is no way to trigger debugging

That's why we need the Debug ID for the page load itself, not for a save attempt. Start debug output logging, load the page, stop logging. If this is intermittent, you'll want to click the "Clear Output" button before each page load to make sure the output covers only a load for which the icon doesn't appear.

You can't generate debug output from Firefox in connector mode, though, so you'd have to close Standalone first. Does this still happen with Standalone closed?

james.k.russell

Closed Zotero, closed Firefox. Reopened Firefox, searched PubMed - problem gone (icon present, download successful). Opened Standalone, closed Firefox, reopened Firefox, searched PubMed, problem reappears (with Standalone open). Repeated this cycle, but enabled debug logging this time around. Generated: D1132270324, which I submitted through Zotero. Seems to have to do with a line length problem or perhaps strange characters in something I frequently trigger in my searches (involving AHA).

adamsmith

could you give us the version numbers of your Firefox, Zotero for Firefox, and Zotero Standalone?

james.k.russell

Zotero Standalone 4.0.11
Zotero for Firefox 4.1.12
Firefox 24.0

dstillman

Those other two errors are unrelated. (The too-long name one would be preventing syncing. Find the item in question and fix the name or delete the item and empty the trash.)

But if this issue is only happening via the connector, we'll need you to either generate real-time debug output from Firefox or, probably easier, try to reproduce this via the Chrome connector (if you have Chrome installed), which unlike Zotero for Firefox's connector mode can generate Debug IDs. The Debug ID from Standalone unfortunately doesn't give us enough info.

james.k.russell

Didn't have Chrome installed, but happy to try to help. Installed, added connector, tried to repeat behavior. So far, the error condition has not appeared. Functionally, from my point of view, this is just as good - better really. I'll just use Chrome instead. I'll let you know if the problem comes back.

Thanks for your help.

james.k.russell

Regarding the other error - I can't find the offending record in either my Standalone database or my online Library. There's nothing currently in the trash. Perhaps it is in queue somewhere? The notice persists in Zotero Standalone.

dstillman

If you're still getting the message, it's definitely there. It wouldn't be online, since the point of the message is that it can't sync a name of that length to the server. You should just need to type a few words from the name in the error message into the Zotero search bar from the root of your library.

james.k.russell

I think I've found it, buried deep in a list of authors, not all shown by default.

Thanks again.

aurimas

Do you know how that reference was imported (you can usually figure it out by looking at the Library Catalog filed)? If it was imported directly from a webpage, can you provide the URL?

james.k.russell

Library Catalog field: NCBI PubMed

Most likely imported using Firefox from PubMed web page, search results.

Extra field: PMID: 23439512

adamsmith

"American Heart Association Council on Quality of Care and Outcomes Research, Emergency Cardiovascular Care Committee, Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation, Council on Clinical Cardiology, and Council on Cardiovascular Surgery and Anesthesia"

yup, that would do it. @aurimas - we should probably do some type of sanity check for these super long institutional authors in PubMed.
Or solve this at the import stage more generally? I don't like Zotero ever importing authors it then can't sync.

james.k.russell

fwiw, your guys dedication, responsiveness and expertise is amazing - thank you so much; I'd be happy to endorse you somehow more publicly. I recommend you to all my colleagues. I support you via George Mason, but it seems awfully indirect - you deserve a lot of credit.

dstillman

we should probably do some type of sanity check for these super long institutional authors in PubMed

This is the first one I've seen that's actually valid. Assuming we actually want creators like this in a single field, we could just increase the length limit a bit on the server side. The only reason I haven't before is that the examples I've seen have always just been concatenated or imported incorrectly.

aurimas

I don't think that's actually a single author. It looks to me like this should be:

• American Heart Association Council on Quality of Care and Outcomes Research, Emergency Cardiovascular Care Committee (idk if this means that the council is part of the committee or if it's actually a different party)


• Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation


• Council on Clinical Cardiology, and Council on Cardiovascular Surgery and Anesthesia

james.k.russell

the long 1st item actually refers to two organizations, not a subcommittee of one, i.e.

American Heart Association Council on Quality of Care and Outcomes Research

and

Emergency Cardiovascular Care Committee (which is also an American Heart Association committee)

aurimas

Correct. And actually, all of those committees are not independent authors per se. The PDF states

Comilla Sasson, Hendrika Meischke, Benjamin S. Abella, Robert A. Berg, Bentley J. Bobrow, Paul S. Chan, Elisabeth Dowling Root, Michele Heisler, Jerrold H. Levy, Mark Link, Frederick Masoudi, Marcus Ong, Michael R. Sayre, John S. Rumsfeld and Thomas D. Rea on behalf of the American Heart Association Council on Quality of Care and Outcomes Research, Emergency Cardiovascular Care Committee, Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation, Council on Clinical Cardiology, and Council on Cardiovascular Surgery and Anesthesia

So it seems that this is more of an author association rather than an independent author. Though maybe not.

adamsmith

thanks.
This is all in one single CollectiveName node in the PubMed data. Not sure what to do here, since they're separated by commas, but there are also commas within organizations.

aurimas

I don't think there is anything we can do. This should be reported to PubMed.

Edit: this is also not a general case on PubMed: http://www.ncbi.nlm.nih.gov/pubmed/22858728 So it definitely seems like a mistake on their end.

Edit 2: As per http://www.nlm.nih.gov/services/pubmederror.html I sent NLM a message regarding this error. Let's see how this process goes.

james.k.russell

From a user's perspective, it would be preferable for you to truncate ridiculously long attributions rather than to import them if it will break synchronization.

aurimas

We would rather give users a chance to correct such errors manually than truncate information. I do think that there probably needs to be a better way to inform users of such errors, but I cannot currently think of a good, non-invasive way to accomplish this.

adamsmith

I think I disagree with that. The sync errors for excessively long authors are a pest & really confusing, I think avoiding those should have high priority.
I don't like throwing out information, but unless we find a good way of handling this (which would be the first best option) I think the second best option is to truncate this on import. If you really want to you can put it into a note. (The author "xyz..." was shortened on import)...

aurimas

If you really want to you can put it into a note. (The author "xyz..." was shortened on import)...

I'd be ok with that. While users should always proof-read Zotero-generated bibliographies, I feel that some mistakes are quite hard to notice (this one may or may not be obvious depending on where a truncation occurs). So I don't feel that comfortable silently importing incorrect metadata. An attached note is a good start, but it still seems somewhat obscure to be readily noticeable.