Cisco's recent redesign of the public Cisco.com website involved integrating the website with a content management system and gathering content from diverse sources.
As content was gathered and imported into the CMS, there was potential for things to get screwed up. Checking each piece of content was a massive undertaking so I helped to design and implement a system that would allow for keeping track of which content that had passed QA, content that had not passed, etc.
Issues that complicated matters:
- The list of content came is a massive export from the CMS. The import mechanism needed to look at each new export as a new snapshot. Information that had changed from one snapshot to the next was considered important.
- Content could be un-checked, checked and approved or checked and unapproved. Additionally, content that had been checked in the past and assigned a status, then changed had a second set of statuses
- The system had to support a manager assigning large groups of content pieces to many QA workers. It had to track and organize content for those workers so that it was easy for them to check content, assign status, and repeat while the application managed and updated their active lists for them.
The system used perl scripts to automate import of the snapshots from the CMS, and access front end for the administrator and QA workers, and a MSSQL backend.