[mythtv] Re: PATCH: mythfilldatabase v1.43 -- arbitrary end times
JTech at interbaun.com
Mon Mar 3 04:45:18 EST 2003
At 3/2/2003 09:10 AM, you wrote:
>On Friday 28 February 2003 07:59 am, J wrote:
> > This patch should eliminate the last of the "unknown" programs, at least
> > for the North American grabber. It scans the program database as a whole
> > and removes any gaps caused by programs that don't end at midnight.
> > In addition to patching filldata.cpp, you must place the file
> > "fix_end_time.sql" in the /usr/local/share/mythtv directory (or,
> > obviously, change the patch)
>While I do like the results, I'm not so sure about the method (the temporary
>tables, etc). Would it be possible to just do the same thing inside of
>mythfilldatabase, and not require the temp tables?
Yes, however when I was experimenting with the algorithm, I found that the
batch file to create the temp databases, was both faster and less memory
intensive. Since this approach needs to have the entire program database
indexed, doing that without the temp databases would mean that the whole
program database would have to be read into memory, indexed, and then,
rather than doing comparisons on the very small subset of entries (in
temp2), you would end up reviewing all the entries. For example, my program
database has 17603 records, but temp2 only contains 465. Since mysql is
already optimized for this sort of operation, it seemed the most efficient
and elegant solution.
That being said, is it the use of the temp tables, the batch file, or both
that you find objectionable?
More information about the mythtv-dev