Opening Session for Mashup U
Some intro to what is mashup – not going to hehash. John Herren mentions that aggregating raw data is a great way to not be creative but still create something cool. By aggregating he means things as simple as sum, mean, … to things like classification, NLP… He mentions the yahoo web service that allows you to send text and it will analyze for important phrases. On to RSS and ATOM – starts with EBay. Has a pyramid with Data feeds on bottom up to APIs then code libraries. GUI Tools, and Platforms.
Does the request-response, SOAP, and REST overviews. He is not a fan of SOAP but says it might be good when you need tight specification on what you want to send and receive. On to AJAX. But then says use a code library for your language of choice. He is a fan of Zend – which is not a suprise given his BIO – but then demos some flickr stuff.
Platforms are ways to host your own mashup on the providers machines – example would be Yahoo! Pipes. He then walks through a Pipes example that takes NYT headlines and finds related YouTube videos.
Browse Based Auth- you can use Yahoo or Google for authentication for your mashup. Allows users to let you access their data without giving you their google or yahoo username and password. BBAuth for Yahoo
Pain Points for mashup – Latency and Scalability, API limits, abuse & metering, API versioing, IP Copyright, Licensing, Reliability & SLA, monetizaion, privacy, security and identity.
Session is done next is SnapLogic.
Pretty standard stuff hopefully shows more interesting stuff to come. That is all the analysis I am gonna do on that session.
Looks to be a service for integrating data sources on the server side. Take data services and bring them together and expose them as REST. Open source and kinda like Pipes but different somehow but he doesn’t explain how it is different. Explains that there is not that much data out there. There are some but not really that much. That’s where they come in – give access to data without a mashup-able api. They let you get data out and in in a lot of different formats throughout the process. They also do “joins” which is much more natural in the mashup framework – not a = but closer to a like.
So their model is that they run server side and allow you to work with a greater variety of data sources. It is kinda a server to provide new mashups feeds. Could be good inside a firewall. They show an example talking to SugarCRM, Quickbooks, and SalesForce.com – do some processing on them, and then expose that as a new service to then be re-exposed as a new feed. The client for defining services and how they fit together is a web client. Now he is live building a demo – I am starting to fade on this one…
Pretty cool that they have built a web UI for doing “database” mapping. You can take two data sources and map fields between the two sources. The do have libraries for doing the work without a GUI – he is showing it in Python.
Side notes about the conference. As more people start signing on to the network the wifi starts to get flakier. It seems rare to find a conference that plans their wifi correctly. There are about 40 people here right now.