add logging for missing events

[Robert Jäschke]

Not all clicks on href's are currently logged. Please check, why and repair it.

[Daniel Zoller]

at least all clicks that are not opened in a new window or new tab aren't logged because the javascript is unloaded before it can make the ajax request to the logging service.

[Daniel Zoller]

i think we should also reconsider to save the "intrahash/user context" of the clicked link instead of saving the position of the list item in the resource list (by the way this is broken). To get the clicked "intrahash/user context" from the list position you have to do the exact same query that was used for the page on that specific date for the requested page. This is not achievable in my point of view!

[Daniel Zoller]

@Nils: please rewrite all external links (like the DOI or URL link of a publication post) to the url /logging/externalurl/<URLENCODED_EXTERNAL_LINK> when the user clicks on an external link (by using JavaScript). /logging/externalurl/<URL> will be redirected by our Proxy (that is not part of your job, only ensure that the browser opens the url /logging/externalurl/<URL> instead of the url <URL>). To identify external links i would suggest to add a rel="external" attribute to all urls that don't point to another BibSonomy site in the jspx and tagx files.

Please ensure that the user can copy the original url (without /logging/externalurl/...) with a right click and "copy url".

[Robert Jäschke]

I am pretty sure that this will not work. Please don't start with that task before we have clarified some issues.

Before we start to rewrite all links, we should discuss this at the BibSonomy Telco. Rewriting breaks several things which I am not in favor of:

1) Letting Copy & Paste still work will most certainly not work (for security reasons) or only with bad tricks (see http://www.heise.de/security/meldung/Tricks-neu-aufgelegt-Vorsicht-bei-Copy-Paste-1841048.html) 2) This breaks the whole idea of how the web works, i.e., linking together resources. 3) Users can not see, if they have visited the link before or not (I know, our CSS breaks this anyway but that's just another argument against our CSS ;-). 4) External sites don't get the proper referer. 5) Google might penalize this (to check). 6) This breaks inline semantic markup (e.g., schema.org, RDFa, etc.) 7) Sites including BibSonomy HTML would require users to access BibSonomy to go the resulting link.

We'll find certainly more issues.

[Stephan Doerfel]

This ist the click-logger discussion. It needs to be resolved very quickly by the seniors

[Daniel Zoller]

by sending the data synchronously