Cleaning things up
So the site in final beta form was released yesterday. The last build attacked various small issues and also allowed us to clean up some of the code and implement some things in a better fashion.
Blog Aggregator
For example, we know have an OPML feed and a blog roll:
The blog roll was very easy to generate since we already have our blog aggregator working. The blog aggregator can process RSS and Atom feeds with some level of flexibility (i.e. recognizes the feed type and processes accordingly).
The aggregator gets all feeds from a Sharepoint list and crawls the feed. It parses them and places the last post of all of them in a single file. Then, a small console app runs every four hours (with a Scheduled Task) that updates this file.
Search
Search was fun to setup and customize. Since our site is in itself some kind of "search" site (in the sense that it concentrates information from other places nicely into one repository), our search goes directly to our Community Server, blogs.officezealot.com.
Some of the last minute challenges we found:
- Anonymous authentication was not working for the results page. The solution was simply to recreate the site. We spent a good amount of time trying to figure out what had happened to no avail. The most likely culprit however, seems to be the master page customization and one of the controls used in search.
- Not enough metadata in pages. Amazingly, most sites are not very good at providing metadata (author, description, taxonomy). Our search relies on crawled properties and managed properties.
- Anonymous authentication for resources outside of Sharepoint. The solution was simply to place these resources under its own application context (this way Sharepoint is not involved in handling the URL).
Code Cleanup
Given the speed at which we created this site, we needed to partition the work in smaller pieces for a team of 4 people (some worked with the master pages, others with webparts and yet another with data migration). In the end, the code looks as if one had created it thanks to code formatting and analysis tools. There is still some work to be done to create a common assemly with some of our helper classes...
Sharepoint quirks
Some of the interesting issues that "just happened" in this project:
- Items in a document library would not update after item.Update() was called: solution, recreate the document library column with the same name. Update works afterwards.
- Permissions. No matter what, setting anonymous access correctly for resources across the board was problematic (but this is probably the case due to some of the complexity of this solution, with many assets, like the Community Server web part outside Sharepoint control).
- Sharepoint Designer errors. With some pages,the editor would freeze. We've found that for complex pages, it is better to use Visual Studio as the editor.
Overall it was a fun and challenging project (time-wise).