Cumulus Support

Hi.

I don't really know if this belongs in questions, bug reports or enhancement requests.

I have just noticed a possible error (bug) in my dayfile.txt.
It has to do with the timestamp around midnight rollover.
It's best seen in the NOAA monthly report.

For example:
Low temp on the 18th is set to 4,5 at 00:00 and on the 17th 6,3 at 06:20.
The only way this could make sense, is that the temp dropped by some 2 deg or more just minutes before midnight, (I use 10min logging interval).
Even thou this is possible, this was not the case this very day.

Examining the log file reveals that the 4,5 00:00 actually belongs to the 19th.

The way I see it, either the 4,5 00:00 record has ended up on the wrong day, or should read 24:00.
Or perhaps left as it is, and the 23:50 figure being used instead.

I see the same thing between the 1st-2nd, 12th-13th and 15th-16th.

Or have I missed something here?

I have a Davis Vantage Vue station and use Cumulus 1.9.4 b1084.
I see the same in 1.9.3.

Regards

Please zip up the diags folder and attach it.

steve wrote:Please zip up the diags folder and attach it.

Here goes....

The minimum temperatures for each day are correct, as far as I can see. Your lowest temperature on the 18th was 4.5 degrees at the very end of the day. When using logger data, it processes the midnight entry before doing the rollover (so that rain is attributed to the correct day). The timestamp on the midnight entry is 00:00.

Yes, it could, if I added the extra code to handle it, use 24:00 for midnight at the end of the day, but it doesn't - sorry. You can call that a bug if you want, but it's not something I can change easily.

steve wrote:The minimum temperatures for each day are correct, as far as I can see. Your lowest temperature on the 18th was 4.5 degrees at the very end of the day. When using logger data, it processes the midnight entry before doing the rollover (so that rain is attributed to the correct day). The timestamp on the midnight entry is 00:00.

Yes, it could, if I added the extra code to handle it, use 24:00 for midnight at the end of the day, but it doesn't - sorry. You can call that a bug if you want, but it's not something I can change easily.

Thanks.

No, don't really want to call this a bug, just as I know how to interpret things.
If I understand You correct, 00:00 can actually mean both beginning of day OR end of day, just consider things in the proper context.

XBower wrote:If I understand You correct, 00:00 can actually mean both beginning of day OR end of day, just consider things in the proper context.

Yes; it's unfortunate as you don't know whether a minimum temperature recorded as being at 00:00 on a particular date occurred at the start or end of a day without looking further at your data, and I agree that a common method of disambiguating is to use 24:00 for midnight, and Cumulus could do this.

If you were not relying on logger data overnight, then I think it is unlikely you would ever see this, as the latest time at the end of the day that could be used would probably be 23:59.

steve wrote:

XBower wrote: If you were not relying on logger data overnight, then I think it is unlikely you would ever see this, as the latest time at the end of the day that could be used would probably be 23:59.

Hi.

Well, for starters I do.
I will probably run the system 24/7 later on.

If I manually change these timestamps to 24:00 (to make the reports look better), would You say it's safe, or could I muck up something else somewhere?
Or, would it be safer to change to the earlier (23:50, I use 10min logging interval) value?
Actually the 23:50 value DOES belong to the day in question, the 00:00 belongs to the NEXT day, in my opinion.

Regards

XBower wrote:If I manually change these timestamps to 24:00 (to make the reports look better), would You say it's safe, or could I muck up something else somewhere?

I think you'll probably get away with it.