I'm not sure if this is the same problem, but it seems closely related... I'm using standalone with Chrome, and when I open a pdf in the Web browser I no longer see the right-click option to "Save Zotero Snapshot from Current Page". These are not proxy server closed sites, but publicly available pdfs (e.g. http://www.affinity.org.uk/downloads/foundations/Foundations%20Archive/37_02.pdf ).
It looks like Chrome displays a different right-click menu when displaying a pdf.
Tom Fish, that's a different problem from the one described in this thread. Assuming you've checked that the connector is installed, try disabling other Chrome extensions (in particular <a href="http://www.zotero.org/support/known_issues#google_chrome_connector">Bookmarks Menu</a>, which is known to be incompatible). If you're still having issues, start a new thread.
Yet another Chrome + Standalone with proxy "situation". This is on a Mac running Snow Leopard. Versions of other software: Firefox 13.0.1, Zotero Standalone 3.0.7, Chrome 20.0.1132.47.
With Firefox, I can do a single click on the Zotero URL-bar icon to get reference metadata and PDF automatically downloaded. I use local storage for PDFs (well, actually a Dropbox directory symlinked to Zotero's "storage" directory).
For some library access, I activate FoxyProxy in Firefox to redirect _all_ http traffic to a remote proxy. Whether the FoxyProxy proxy is active or not, the metadata and the PDF download (as reflected by metadata in Zotero Web and metadata+PDF in Standalone Zotero). Brilliant!
With Chrome, the same thing works when I'm not using the proxy: single click on the Zotero URL-bar downloads metadata to Zotero Web and metadata + PDF to Standalone. Chrome > Preferences > Extensions > Zotero Connector > Options claims that "Zotero Standalone is currently available."
When I fire up Chrome to use a proxy (using a commandline like: "open -a '/Applications/Google Chrome.app' --args --restore-last-session=5 --proxy-server=my.proxy.server.org:80"), things semi-break. Only metadata get downloaded, and only to the web database (I can see them in Standalone by clicking Standalone's refresh). No PDFs get downloaded. The Chrome extension preference for Zotero claims "Zotero Standalone is currently unavailable" (the "unavailable" becomes set the moment I validate to the remote proxy in the browser using Basic Auth).
Note that this is all independent of clever connectors or magical autoproxying. This is using a bog-standard full-on proxy-everything. It doesn't matter whether the site is "open" (e.g. PLoS ONE) or protected (e.g. http://www.int-res.com/abstracts/meps/v458), though of course without an active proxy, I get no PDFs from protected sites under any setting.
It looks as though the communication between the Chrome Zotero Connector and Zotero Standalone gets snapped when Chrome runs using a proxy.
And it gets us halfway there: the browser now "sees" the Standalone Zotero (it now claims "Zotero Standalone is currently available") and metadata get immediately posted to the Standalone copy. (Just for kicks I also tried "localhost" — that fails outright.)
Still missing: PDF download from otherwise-protected sites. For example, on the Marine Ecology Progress Series site, "open" PDFs get downloaded but non-open ones don't (when the proxy is enabled).
Speculating here: I'm assuming that the way things work is that the Zotero Connector signals Standalone to do the metadata-fetch and the PDF-fetch. So now the connection allows Zotero Standalone to grab the metadata (but the metadata are available proxy-or-not), but Zotero Standalone fails to grab the PDF since it doesn't "know" about the proxy. Since only the browser speaks through the proxy, Standalone gets only the metadata.
So... elaborate the protocol between the two so that Connector could tell Standalone about the proxy? Or is the proxy information even available (explicitly) to the Connector inside Chrome?
If you don't want to change your system proxy settings, it is possible to change the proxy settings in about:config. To get to about:config, open the Zotero preferences, click "Advanced" then "Open about:config". You probably want to set network.proxy.type to 1 and set network.proxy.http and network.proxy.http_port to the hostname and port of your proxy. See MDN for more documentation about proxy settings. If your proxy requires authentication, this may or may not work.
I think it's now possible to get information about the proxy from the connector and send it back to Zotero Standalone, but the proxy configuration you have seems relatively uncommon, and it may be a little while before I can get to this.
Yes, you're right: I don't want to change system proxy settings (in fact, for many of our users, they don't have the client rights to change system proxy settings).
Doing the changes you suggest doesn't (quite) work. Same results: metadata download fine, PDF does not. Sniffing the wire traffic shows that the problem appears to be the password request by the proxy. Since Zotero Standalone doesn't/can't reply, the proxy comes back with a "407 Proxy Authentication Required" error instead of the requested PDF.
How often have you wished that proxies could just go away, or at least that everyone would stop making clever innovative versions of them that you have to try to accomodate?
Thanks for the suggestions — would love a solution, but you're right, this is probably a pretty unusual configuration. Though being able to respond to a Basic user/password request may not be unusual.
Followup to my own posting: Having put in the network.proxy.http and network.proxy.http_port settings, then seeing Zotero Standalone fail to fetch a PDF since it failed to get a user/password validation to the proxy server, I stopped.
However, I subsequently dropped a PDF into Standalone manually, then did a right-click > Retrieve Metadata for PDF. That popped up a user/password dialog box for the proxy, and subsequent retrieval attempts through the browser work (yay!).
My take on this is that direct interaction with a PDF (or perhaps anything else that sends it out to the web) in Zotero Standalone will trigger a user/password dialog for a proxy that Standalone has been configured to use. Having the browser signal Zotero Standalone to fetch a PDF, however, fails to trigger the user/password dialog box.
So... is there a way to make proxy user/password dialogs happen when Standalone is triggered by a browser? Or do I have to keep an un-metadata-ed PDF hanging around on my desktop to "prime" Standalone's proxy usage?
Upgrade Storage
It looks like Chrome displays a different right-click menu when displaying a pdf.
With Firefox, I can do a single click on the Zotero URL-bar icon to get reference metadata and PDF automatically downloaded. I use local storage for PDFs (well, actually a Dropbox directory symlinked to Zotero's "storage" directory).
For some library access, I activate FoxyProxy in Firefox to redirect _all_ http traffic to a remote proxy. Whether the FoxyProxy proxy is active or not, the metadata and the PDF download (as reflected by metadata in Zotero Web and metadata+PDF in Standalone Zotero). Brilliant!
With Chrome, the same thing works when I'm not using the proxy: single click on the Zotero URL-bar downloads metadata to Zotero Web and metadata + PDF to Standalone. Chrome > Preferences > Extensions > Zotero Connector > Options claims that "Zotero Standalone is currently available."
When I fire up Chrome to use a proxy (using a commandline like: "open -a '/Applications/Google Chrome.app' --args --restore-last-session=5 --proxy-server=my.proxy.server.org:80"), things semi-break. Only metadata get downloaded, and only to the web database (I can see them in Standalone by clicking Standalone's refresh). No PDFs get downloaded. The Chrome extension preference for Zotero claims "Zotero Standalone is currently unavailable" (the "unavailable" becomes set the moment I validate to the remote proxy in the browser using Basic Auth).
Note that this is all independent of clever connectors or magical autoproxying. This is using a bog-standard full-on proxy-everything. It doesn't matter whether the site is "open" (e.g. PLoS ONE) or protected (e.g. http://www.int-res.com/abstracts/meps/v458), though of course without an active proxy, I get no PDFs from protected sites under any setting.
It looks as though the communication between the Chrome Zotero Connector and Zotero Standalone gets snapped when Chrome runs using a proxy.
Any suggestions (besides "use Firefox":)? Thanks!
And it gets us halfway there: the browser now "sees" the Standalone Zotero (it now claims "Zotero Standalone is currently available") and metadata get immediately posted to the Standalone copy. (Just for kicks I also tried "localhost" — that fails outright.)
Still missing: PDF download from otherwise-protected sites. For example, on the Marine Ecology Progress Series site, "open" PDFs get downloaded but non-open ones don't (when the proxy is enabled).
Speculating here: I'm assuming that the way things work is that the Zotero Connector signals Standalone to do the metadata-fetch and the PDF-fetch. So now the connection allows Zotero Standalone to grab the metadata (but the metadata are available proxy-or-not), but Zotero Standalone fails to grab the PDF since it doesn't "know" about the proxy. Since only the browser speaks through the proxy, Standalone gets only the metadata.
So... elaborate the protocol between the two so that Connector could tell Standalone about the proxy? Or is the proxy information even available (explicitly) to the Connector inside Chrome?
I think it's now possible to get information about the proxy from the connector and send it back to Zotero Standalone, but the proxy configuration you have seems relatively uncommon, and it may be a little while before I can get to this.
Doing the changes you suggest doesn't (quite) work. Same results: metadata download fine, PDF does not. Sniffing the wire traffic shows that the problem appears to be the password request by the proxy. Since Zotero Standalone doesn't/can't reply, the proxy comes back with a "407 Proxy Authentication Required" error instead of the requested PDF.
How often have you wished that proxies could just go away, or at least that everyone would stop making clever innovative versions of them that you have to try to accomodate?
Thanks for the suggestions — would love a solution, but you're right, this is probably a pretty unusual configuration. Though being able to respond to a Basic user/password request may not be unusual.
However, I subsequently dropped a PDF into Standalone manually, then did a right-click > Retrieve Metadata for PDF. That popped up a user/password dialog box for the proxy, and subsequent retrieval attempts through the browser work (yay!).
My take on this is that direct interaction with a PDF (or perhaps anything else that sends it out to the web) in Zotero Standalone will trigger a user/password dialog for a proxy that Standalone has been configured to use. Having the browser signal Zotero Standalone to fetch a PDF, however, fails to trigger the user/password dialog box.
So... is there a way to make proxy user/password dialogs happen when Standalone is triggered by a browser? Or do I have to keep an un-metadata-ed PDF hanging around on my desktop to "prime" Standalone's proxy usage?
Cheers!