This is an old concept and originated in the days when we thought most Endeca applications would have a single 'primary' dimension, such as a Category - since the early applications all had something like this, we knew that every record would have at least one assignment to this dimension and we could use it as a shortcut for figuring out the total number of records. Of course the apps that people build today don't always have a conceptual primary dimension, and there are other, less convoluted, ways we can figure out how many records are in the system, so the concept is a little out of date.

We took the first step of masking it for baseline updates several releases ago (since dgidx creates and assigns it to each record for you automatically), but partial updates bypass dgidx completely and we didn't make the changes in forge or dgraph -- therefore, you have to manually assign the primary dimension to each record in a partial update, and to do this you need to know about the primary dimension ahead of time (this is why the option is still exposed in Dev Studio). Removing the primary dimension concept completely is something we plan for a future release, since there really isn't something you can do with the primary dimension that you can't do with any regular dimension.

'Dev > endeca' 카테고리의 다른 글

Interaction with Operational Environment  (0) 2010.03.09
dimension 과 autogen  (0) 2009.12.15
Utility 'emgr_update_get_ws_settings' failed.  (0) 2009.09.29
start_endeca_http_service.sh  (0) 2009.09.28
File exclude expressions (Crawling)  (0) 2009.08.26

+ Recent posts