4063 - Solar chart missing

[Ulric2]

After installing 4063 (over 4043) on Ubuntu, the Recent Solar chart does not work - just says "Loading" but never loads. It worked just fine in 4043.

On the bright side, the .ini file survived!

[broadstairs]

I just checked mine and it opened fine, do all the other charts open OK? Any errors in log file?

Stuart

[Ulric2]

I have investigated and I think I've found the reason.
I have no instruments connected for solar or UV because radio interference I receive disrupts the readings. I analyse this interference though, for other reasons. It seems that 4063 places NULL in the CumulusMX.db file in these fields where 4043 put 0. I guess a NULL in the column stops the chart from working.

The simple answer is that I need 0 rather than NULL in these columns. How can I achieve that?

[broadstairs]

If you have no solar sensor available why not turn off all solar in Settings/display options? That way the graph should not show. Why do you need zeros in the data?

Stuart

[Ulric2]

Confirmed. Updating the recent data table to coalesce the nulls to zero fixes the problem and the chart displays normally. As soon as cumulus writes a NULL to the table (every minute) the chart breaks.

[Ulric2]

If you have no solar sensor available why not turn off all solar in Settings/display options? That way the graph should not show. Why do you need zeros in the data?

Stuart

Solar signals caused by radio interference are received sporadically. They do not depend on the presence of the instrument. I need these signals displayed on the chart. If there is a NULL anywhere in the data, the chart never loads. Could the query that drives the chart just do COALESCE(SolarRad,0) SolarRad, COALESCE(UV,0.0) UV ? That would fix it immediately.

[broadstairs]

I am confused, if you have no solar sensor how does CMX receive this data and know how to save it?

Stuart

[mcrossley]

Simple answer you cannot. MX is moving towards using null values for data that is "missing".

But I can fix the code that generates the graph files - fixed for version 4.3.1

[Ulric2]

I am confused, if you have no solar sensor how does CMX receive this data and know how to save it?

Stuart

It is radio interference which affects the base station. Essentially, the base station receives a spurious signal (generally SolarRad = 1662, UV = 2) from an illegal device being operated in the vicinity. It appears as spikes in the Solar and UV data. The timings of these spikes are essential to another project of mine and the solar chart in MX is the easiest and best way to view them!

[Ulric2]

Simple answer you cannot. MX is moving towards using null values for data that is "missing".

But I can fix the code that generates the graph files - fixed for version 4.3.1

Thank you! That is good news.
Could it be that the problem will also exist in the Daily Solar Chart?

A horrible thought crosses my mind. Will cumulus now ignore SolarRad and UV if the sensors are not detected at startup? That could be a problem.

[Ulric2]

Also seems to have stopped the Solar and UV traces at Wunderground dead in their tracks.

https://www.wunderground.com/dashboard/pws/IBALDO9

If you examine historic months there, you will see the signals I'm trying to record and track.

[mcrossley]

A horrible thought crosses my mind. Will cumulus now ignore SolarRad and UV if the sensors are not detected at startup? That could be a problem.

No, it will just record null values until some data is received.

[Ulric2]

No, it will just record null values until some data is received.

Also good news. For the moment, I have coalesced the values when they are transferred up to my SQL Server so no problems there.

I look forward to the return of the graphs in 4.3.1 .

[Ulric2]

Another thought - I assume you're not writing NULLs to the Dayfiles?

EDIT: Yes, MX is writing an empty string "" to the DayFile where previous versions wrote "0". This caused one of my Python scripts to fail but it was an easy fix.

[mcrossley]

Going forward more data values will start support null. The present system of using zero for missing data is not optimal.