Page 2 of 3

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 3:05 pm
by admin
The just released update of Meteobridge does now send pressure data via "baromin" parameter. As worked out by user gus the "pressure" parameter expects Pa, which is rather unusual, and I did send hPa as other networks ask for. That explains factor 100 too low readings. So it is not an error on the windy side, just an unusual setting and I did not read closely enough. Anyway, I hope latest version is working now fine.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 5:46 pm
by gus
admin wrote:
Thu Apr 18, 2019 3:05 pm
The just released update of Meteobridge does now send pressure data via "baromin" parameter. As worked out by user gus the "pressure" parameter expects Pa, which is rather unusual, and I did send hPa as other networks ask for. That explains factor 100 too low readings. So it is not an error on the windy side, just an unusual setting and I did not read closely enough. Anyway, I hope latest version is working now fine.
Thank you!

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 8:34 pm
by lightmaster
gus wrote:
Wed Apr 17, 2019 2:05 am
I thought about the time stamp but it seemed like more work for not much benefit. Although I admit to not really know how Retry and lost connectivity really work. For this reason I suggested setting Retry to 2. That equals 5 minutes. After that point the data if it was not received is useless as there is new data to be had for the next 5 minute cycle. So why not attempt to upload that new data instead of keeping the attempt to upload what would be now 5 minute old data? Unless the Meteobridge buffers the old data with the time stamp and say it goes after say 7 minutes for example. I've seen my Meteobridge show "Buffer: X items" in the System tab but I haven't read the documentation to really understand it. What kind of stress might I put the Meteobridge into if I start accumulating too many buffered items? For this reason all my upload services are set to 2 Retries or 5 Retries depending in if I'm uploading to that service every 5 or 10 minutes. (5 Retries is 12 minutes). Which is why I don't understand why I would ever want to choose unlimited Retries...as that sounds like a sure way to drown the Meteobridge.
Unlimited definitely would end up drowning the Meteobridge and it would take forever to get caught up. From my experience with WU and having it set to 10 retries, if something happens and data can't get out, it keeps retrying that old data while also queueing new data as well. Sites like WU that accept old data with timestamps will save that data and attribute it to the proper time per the timestamp. So if you internet goes out for 10 mins, that data will eventually end up saved on their site under the correct time.

Downside being that it seems the Meteobridge sends data serially, so if WU had 500 queued datas (wtf is the plural of data?), it would be some time before WU saw the latest data. So setting a reasonable retry limit is definitely necessary. That being said, you set the retry to 5 and include the timestamp, the queue doesn't get bogged down and the site gets the old data and knows when the measurements were taken.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 9:23 pm
by johnbowe
Nearly there, now a decimal point needed before the last digit in the pressure reading, or ditch the last digit.

Windy Readings 2.JPG
Windy Readings 2.JPG (18.61 KiB) Viewed 529 times

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 9:38 pm
by gus
lightmaster wrote:
Thu Apr 18, 2019 8:34 pm
gus wrote:
Wed Apr 17, 2019 2:05 am
I thought about the time stamp but it seemed like more work for not much benefit. Although I admit to not really know how Retry and lost connectivity really work. For this reason I suggested setting Retry to 2. That equals 5 minutes. After that point the data if it was not received is useless as there is new data to be had for the next 5 minute cycle. So why not attempt to upload that new data instead of keeping the attempt to upload what would be now 5 minute old data? Unless the Meteobridge buffers the old data with the time stamp and say it goes after say 7 minutes for example. I've seen my Meteobridge show "Buffer: X items" in the System tab but I haven't read the documentation to really understand it. What kind of stress might I put the Meteobridge into if I start accumulating too many buffered items? For this reason all my upload services are set to 2 Retries or 5 Retries depending in if I'm uploading to that service every 5 or 10 minutes. (5 Retries is 12 minutes). Which is why I don't understand why I would ever want to choose unlimited Retries...as that sounds like a sure way to drown the Meteobridge.
Unlimited definitely would end up drowning the Meteobridge and it would take forever to get caught up. From my experience with WU and having it set to 10 retries, if something happens and data can't get out, it keeps retrying that old data while also queueing new data as well. Sites like WU that accept old data with timestamps will save that data and attribute it to the proper time per the timestamp. So if you internet goes out for 10 mins, that data will eventually end up saved on their site under the correct time.

Downside being that it seems the Meteobridge sends data serially, so if WU had 500 queued datas (wtf is the plural of data?), it would be some time before WU saw the latest data. So setting a reasonable retry limit is definitely necessary. That being said, you set the retry to 5 and include the timestamp, the queue doesn't get bogged down and the site gets the old data and knows when the measurements were taken.
Great response. Makes sense. Thank you. I guess using the time stamp makes sense for sites that let you upload historical data.

BTW the "data" is already in plural form. Singular form is datum. In your use case you'd want to add plural clarification as to what data is plural. So you could say data sets, data records, or data uploads, and that would have accomplished what you wanted to say.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Thu Apr 18, 2019 9:46 pm
by Andyk1
johnbowe wrote:
Thu Apr 18, 2019 9:23 pm
Nearly there, now a decimal point needed before the last digit in the pressure reading, or ditch the last digit.


Windy Readings 2.JPG
Not sure if it was first but it was not from lack of trying. Boris did a he'll of a job getting it right. Mine now works through meteobridge and wasting time trying to figure out why I could not get it working through the meteobridge HTTP method. I believe whom ever said it was because of different sensor readout is true as I can get it working by not including the UV sensor. Strange but that's the only way I could get temp, hum, press, wind and rain to work. All is good now. Thanks Boris.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Fri Apr 19, 2019 12:28 am
by lightmaster
gus wrote:
Thu Apr 18, 2019 9:38 pm
Great response. Makes sense. Thank you. I guess using the time stamp makes sense for sites that let you upload historical data.

BTW the "data" is already in plural form. Singular form is datum. In your use case you'd want to add plural clarification as to what data is plural. So you could say data sets, data records, or data uploads, and that would have accomplished what you wanted to say.
Assuming Windy is like WU and can receive historical data, might be worth looking into making a script to convert and transmit historical data for your station so you can fill in past years. Would largely depend on any API limits because if it takes 4 month to transmit the data for a year.... I'm not sure I'd trust a script not to crash out during that time frame, lol.

And that's what I get for all but flunking high school Latin, not recognizing that data ends in an -a and is there for plural. If I remember correctly, and I probably don't lol, ending in -um makes it masculine, so data/datum is male?

Nope, I don't remember correctly, ending in -um/-a makes it neutral, not masculine.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Fri Apr 19, 2019 1:18 am
by Andyk1
Hi Boris. Someone jus let me know of a misplaced period in the pressure reading in the Windy.com update addition you made today incase your not aware. Please see screenshot.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Fri Apr 19, 2019 9:37 pm
by BigAlT
I discovered that in obtaining a Windy.com api be careful if you’re just selecting the api in the window and copying and pasting it. The actual api is over twice as long as the window and you may only be copying part of it which will obviously not result in data uploads. On an iPad, highlighting the api in the window doesn’t actually select the whole api. I needed to put my finger on the highlighted window area and pull it to the right to highlight the whole number. The copy button next to the window doesn’t copy the api at all on an iPad, it apparently does work on a laptop/desktop.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Sat Apr 20, 2019 1:32 am
by admin
Just released update should do the pressure finally right. Sorry for the bugs. As I don't use windy myself right now, I was dependent on your feedback.

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Sat Apr 20, 2019 2:00 am
by Andyk1
admin wrote:
Sat Apr 20, 2019 1:32 am
Just released update should do the pressure finally right. Sorry for the bugs. As I don't use windy myself right now, I was dependent on your feedback.
Not a problem. Most went back to the HTTP version by Galfert which worked. We didn't know if the error was on MB or Windy's side. Thanks for all your efforts.

Andy

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Tue Apr 23, 2019 3:37 pm
by johnbowe
Great all seems to be working now (Windy in Weather Nets). I know this must seem obvious, but just wanted to check, in the image below under wind speed, the g:9 indicates the Gust speed?

Windy Readings 3.JPG
Windy Readings 3.JPG (18.74 KiB) Viewed 445 times

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Tue Apr 23, 2019 4:15 pm
by gus
johnbowe wrote:
Tue Apr 23, 2019 3:37 pm
Great all seems to be working now (Windy in Weather Nets). I know this must seem obvious, but just wanted to check, in the image below under wind speed, the g:9 indicates the Gust speed?
Yes g:9 is your gusts for that moment that you have selected on the time scale. Also the green and blue marks indicated are winds and gusts also. Blue is less intense wind and green is stronger winds.

Windy Winds Data.png
Windy Winds Data.png (36.96 KiB) Viewed 443 times

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Tue Apr 23, 2019 5:54 pm
by johnbowe
Cool, thank you Gus

Re: Windy.com seeking input on developing API - Let's make Meteobridge first

Posted: Tue Apr 23, 2019 6:10 pm
by gus
Windy now accepts pressure in hPa (milibar).
  • mbar - real number [milibar, hPa]; atmospheric pressure alternative
https://community.windy.com/topic/8168/ ... o-windy/85