[mythtv] manual record, unset autoexpire, recorded OK, program deleted.

[Jay Jarvinen]

Synopsis: Duplicate entries in 'record' table, manually recorded show
set not-to-expire, got nuked unexpectedly.

I run CVS myth, last update was 2005-02-27 05:19:02 -0600.

Mom called and informed me "Hey, your second cousin will be on ESPN2 in
a few minutes.."

So I set loaded up the program guide, found the program, "Record only
this showing", storage options, "Don't allow auto expire", [go back],
[Save these settings]. (My default setting is to allow autoexpire)

Guide & CPU load indicated the recording had started.

Later, not seeing it in recorded programs, I finally re-found it in the
program guide marked "Single Record - Deleted", press <enter> reveals:

"This showing was recorded but was deleted before recording was
completed."

Press 'i', storage options, (now) shows "Allow auto expire".

When I checked the database, I noticed two entries for the program:
+----------+------------+------------+-----------+----------+--------+
| recordid | autoexpire | startdate  | starttime | endtime  | chanid |
+----------+------------+------------+-----------+----------+--------+
|       91 |          1 | 2005-03-25 | 15:00:00  | 16:00:00 |   1063 |
|       92 |          0 | 2005-03-25 | 15:00:00  | 16:00:00 |   1063 |
+----------+------------+------------+-----------+----------+--------+

Evidently, "it" (guide/autoexpire code) is only seeing the first entry?

The program was already in progress, but it did indeed finish recording
(based on log). At the time, there was definitely room for the show, but
it might have pushed it over the auto-expire free space threshold. It's
busy recording pre-scheduled as I type this.

The rest of the programs in mythconverg.record, only have a single row.

In any case, it clearly didn't do what I expected. If it's user error,
then anytime I'm near or under the free space threshold, I must globally
turn off autoexpire before selecting manual recordings, and manually
ensure there's adequate space.

Bug? Should there ever be duplicate entries in the record table?

I'm hoping someone more capable can answer my questions. I got as far as
grepping the i18n directory, only to find the error message as part of
large array of messages with little context as to where they're used.

Happy to provide more details on request, thanks,

-Jay