So I recently stumbled across this video I had uploaded to Youtube a few years back. I had completely forgotten about it.

This was in the Jive 3.0 or 4.0 days, I believe.


So this was a document browser for Jive, I wrote using an awesome, but sadly mostly forgotten technology from Microsoft, called Pivot (Microsoft Live Labs Pivot - Wikipedia, the free encyclopedia). Pivot was based on the same technology as Photosynth, and allowed you to do some amazing things with large datasets. It allowed you to start with datasets of thousands or even ten thousands of elements and drill down into or filter them on the fly, using a positively Minority Reportesque UI.


While this is not really that relevant from todays point of view, I still think that maybe, just maybe, this little project of mine influenced the cards-tiles interface for browsing people, places and content in Jive 5 and beyond a tiny bit.


Watch the video below. It starts a little slow, but gets a lot shinier around the 3 minute mark. I've provided some technical details below.



From a technology point of view this was pretty involved, due to the lack of decent APIs on the Jive side at that point. It shows how far Jive has come when it comes to APIs.


The process went as follows:

  • Information about content was retrieved directly from the Jive database. This included titles, content, authors, tags, ratings and the place a content was stored in. This was pretty easy and fast. The largest dataset I've tried this with was around 3000 documents.
  • Any binary content was downloaded using the legacy XML based REST API that was available in Jive 3.0 and 4.0. Openclient, the first version of the current API came in 4.5. The reason for this was that the binary content is encoded in some weird undocumented way in the database.
  • Jive native content was converted to images using a tool that came with the Pivot SDK. Basically you defined an HTML template, pumped your Jive HTML content into it, and it would render it out as an image file that Pivot could use.
  • Binary content was converted into images using the incredibly awesome Apose libraries. They allow you to take any office format and convert it into a pixel perfect image or PDF. I wish Jive would use them instead of their OpenOffice solution. Sometimes when Jive has butchered another one of my Powerpoints, I dream of writing a replacement document rendering service using Aspose ;-) We are using their technology in our SmarterPath LMS for Jive and are very, very happy with it.
  • All of the data was mangled into a Pivot collection.

This was certainly done in a batch way. Processing a 1000 document dataset took 5 to 10 minutes on a laptop. The majority of the time converting HTML and Binaries to images.


Being a Silverlight based technology, Pivot isn't going anywhere. Pretty sad, in my opinion. Imaging the Jive people browser allowing you to drill down from 10.000 colleagues using faceted search in real time, without any browsing or paging. *Sigh*


I hope you liked this little trip down nostalgia lane. I have a few more silly examples like this, that I might post. If you don't see any value in this, let me know also.