ClimateDataCalculation

[Mastiff]

I'll let it run a few days more and see what happens. I will be going down to the cabin Wednesday. Oh, and I have to put the wind sensor a bit higher up, it's more windy there than it shows!

[krambriw]

I wonder,,,my memory consumption is going up, please check yours as well, maybe I have to look for new bugs

The memory consumption seems to be dependent on the amount of data in the reports, I assume this is normal and expected and EG is supposed to release it to the system when no longer needed. So it could be just simply that some references variables needs to be deleted

[Mastiff]

Yeah, I see that it's up. Not hysterically, lik when the thread system was worked on, but still... Around 227 000 kb, but it doesn't seem to go higher. It's more like "Professorn":

Det går hit, det går dit, det går runt en liten bit,
och den startar p├Ñ ett kick, det ├ñr en makal├Âs manick.

So a bit up and down, but no real growth from 227 000.

[krambriw]

Thanks, as long as it is stable, no worries

[Mastiff]

Well, it's 228 000-231 000 now, but it's not looking too bad. I'll check tomorrow morning.

[krambriw]

I made a change. Obviously my ambition to update table names regularly was not working perfect from memory usage perspective. So I changed that a bit and added an action that allows you to refresh table names. This might be useful in situations where you have added data capturing of new sensors and jump to reports without re-starting EG in between. In those maybe rare cases you can run the refresh table names action and your report templates will be updated.

New version uploaded

[Mastiff]

Sorry I fell out there . A couple of chaotic days. I'm at the cabin now, and the previuos version had been running since I started it. It was still only at 250 000 kb memory. With the new version it started at 52 000, so quite a change! I'll let you know if it runs in to trouble.

[Mastiff]

The memory seem to stabilize around 70 000.

Another thing: I looked in the file log, and there are now NW (northwest) in there. Only NNW and WNW. So it seems there's a blind spot either in the wind sensor or the plug-in for the RFXtrx. I have tried to add the wind sensor to the NetHomeServer, but that doesn't see it. I read that the support for WGR800 is "theoretically". I can add a question for it there.

[krambriw]

So it seems there's a blind spot either in the wind sensor or the plug-in for the RFXtrx

Good point, but we do have tools to check a bit further...

In the RFXtrx plugin, there is an action called 'decode_Test_Message'. What this thing does is that it injects a message into the plugin where it would appear when it comes from the HW device itself. In order to try if the plugin handles this correctly, we need to inject a message that says wind direction NW. If this is understod/decoded correctly by the plugin, I think the problem is 'outside' so to say.

This message (as it comes from the RFXtrx) is exactly such a message:

Code: Select all

105602020d000145001d00156874205249

If you try and inject it, the following event will be presented in the EG log

Skärmklipp.PNG (3.51 KiB) Viewed 4585 times

[Mastiff]

I'm afraid I don't know how to inject. I may get something from the ONH setup, though. It didn't work now, but maybe later today, if the nightly build isn't put up.

[krambriw]

Aha, I see, sorry, did not explain good enough

1) In the computer running the RFXtrx plugin, add the action 'decode_Test_Message' and the following typical view will be presented

2) What you see is the typical hex code message structure that we receive from the RFXtrx


3) Copy & paste the hex code example I posted above, click Apply and then Test. You will then see an event created, in this example as if it came from the wind sensor

[Mastiff]

Yep, I get northwest from that. I'll have to see what happens when/if I get the ONH/Tellstick to decode the wind sensor.

[Mastiff]

Annother thing: Would it be possible to make the graph lines disappear instead of defaulting to 0 when there are dropouts? I have some dropouts every now and then in some rooms more than others, and that creates a thin line down to zero and then up again. Also, I think it's more logical to have no line at all then a flat line at 0 when there are no signals. If possible.

[krambriw]

My original idea was to fill the gaps with last known values but maybe this is not working as expected
Best I think would be if the gaps could be visualized better somehow maybe in different colour but currently I dont know if highcharts has a built in funtion to handle gaps in data series
Somehow I have to fill in some data also for the gaps, otherwise the positioning of the graph will be wrong from the timeline perspective
Next week I will be back and can investigate this better

[Mastiff]

I see what you mean. If it supports different colors, maybe you could set the gaps to white? That way it will be invisible and not look strange! But it's of course no hurry, so take your time until you get back to where ever you are and what ever you are doing!