danielk at cuymedia.net
Fri May 26 15:41:46 UTC 2006
On Fri, 2006-05-26 at 10:13 -0400, Yeasah Pell wrote:
> Emboldened by Daniel's semi-endorsement of hierarchical diseqc devices,
> I cranked out some fancy tree-based diseqc code last night. It involves
> some DB changes of course, and I'm not sure how that is generally
> handled (obviously if such a patch is accepted a schema version needs to
> be assigned, a schema upgrade needs to be added to dbcheck.cpp, and I
> assume a requirement is that everything be backwards-compatible -- i.e.
> no removing anything from the DB, just deprecation)
Yeah, we normally put the SQL to delete the columns in a comment
at the end of UpgradeTVDatabaseSchema along with the version it
is safe to do the delete (i.e. the version after the next release).
> I figure it could use some going over anyway, so I'm posting the rough
> schema changes that the tree-based code uses currently (with some
> explanation). Essentially there are two new tables -- a table called
> "diseqc" (probably should be called diseqcsetup or something) which
> contains id/value pairs forming a particular configuration of a
> particular diseqc tree (these are associated with card inputs), and a
> table called "diseqctree" which contains the actual tree of diseqc
> devices (these are associated with capture cards)
How about "dtv_device_config" and "dtv_device_tree" ?
Implement the configuration UI before you get too attached
to the schema. I think it is probably easiest to just have
parent pointers in the device tree, with zeroed parent pointers
for the roots.
More information about the mythtv-dev