Hello all,
I recently graduated from the Indiana University of PA with degrees in computer science and mathematics. I am starting grad school in the fall at Binghamton University. I am very much looking forward to my participation in the GSOC 2013. My original proposal follows:
Summary: Improved database layout - The existing SQL database layout for music and videos is steeped in history, not all of it good. The layout is non-optimal for the storage system that XBMC needs for data. The need is for a fast store from which movie information can be retrieved. It's mostly static information, and there's a lot of interlinked information (m:n maps) that need to be associated with each item and displayed when a list is fetched. i.e. either the information is grabbed in a single query, whereby you have a bunch of unnormalised data (e.g. a list of genres for each movie item) or the information is grabbed through multiple queries per item, whereby you can have everything nicely normalised. A hybrid solution might be best - some work has been started on this as part of garbear's piccture library project. Various other suggestions and notes are available on the forums and wiki.
How will I achieve this: The forum suggestions linked to in the proposal seem like the best starting point and would accomplish many of the improvements desired. A particular benefit of this is it allows XBMC to continue using a SQLite database while greatly increasing the databases ability to handle sharing the data to other clients.
Goals:
API methods to: define a type of artwork, define a type of media (Home Movies, etc.), add movies to one or more sets
Get a working version of Frodo using DecK's database idea (or a modified version thereof)
After some discussion with members of Team XBMC, I am going to be working on a stand-alone application, which will act as a proof-of-concept and testbed for the new database and corresponding API.
I recently graduated from the Indiana University of PA with degrees in computer science and mathematics. I am starting grad school in the fall at Binghamton University. I am very much looking forward to my participation in the GSOC 2013. My original proposal follows:
Summary: Improved database layout - The existing SQL database layout for music and videos is steeped in history, not all of it good. The layout is non-optimal for the storage system that XBMC needs for data. The need is for a fast store from which movie information can be retrieved. It's mostly static information, and there's a lot of interlinked information (m:n maps) that need to be associated with each item and displayed when a list is fetched. i.e. either the information is grabbed in a single query, whereby you have a bunch of unnormalised data (e.g. a list of genres for each movie item) or the information is grabbed through multiple queries per item, whereby you can have everything nicely normalised. A hybrid solution might be best - some work has been started on this as part of garbear's piccture library project. Various other suggestions and notes are available on the forums and wiki.
How will I achieve this: The forum suggestions linked to in the proposal seem like the best starting point and would accomplish many of the improvements desired. A particular benefit of this is it allows XBMC to continue using a SQLite database while greatly increasing the databases ability to handle sharing the data to other clients.
Goals:
API methods to: define a type of artwork, define a type of media (Home Movies, etc.), add movies to one or more sets
Get a working version of Frodo using DecK's database idea (or a modified version thereof)
After some discussion with members of Team XBMC, I am going to be working on a stand-alone application, which will act as a proof-of-concept and testbed for the new database and corresponding API.
