Build 958 Crash

[nking]

Just found that Cumulus has crashed, elf attached.

[steve]

It looks like it had an error writing dayfile.txt and hence the date wasn't set up, so it then crashed - I'll change the code so that doesn't happen again. Have a look in the diags log to see if there is an error about dayfile.txt at midnight.

[nking]

Have a look in the diags log to see if there is an error about dayfile.txt at midnight.

Yes there is. I assume the data will be fine and I can continue as is?

[steve]

You probably have an entry missing in dayfile.txt for yesterday.

There seems to be a bug in 958 which causes the failure to write to dayfile.txt if it's done the 'find the initial monthly/yearly highs and lows' thing in the same run.

[Synewave]

I too have just had Cumulus crash.

[steve]

I too have just had Cumulus crash.

Same problem.

[nking]

Steve,

I made a copy of the complete Cumulus directory so I think I will revert back to this (build 957) for the moment and wait for your next build, it will save me having to go through my data correcting any errors.

[steve]

The crash occurs because the code processing the dayfile.txt to find the highs and lows encounters some kind of problem and forgets to close the file. I'll have a new build available shortly.

[steve]

As an aside, I've found a few other places where the same problem might occur when dayfile.txt has 'corruptions', possibly explaining the I/O errors some people get which I have been blaming on anti-virus etc

[nking]

Steve,

I've just installed 959.... very quick update

You have an amazing work ethic, It must be that you never have any free time to just read a book or watch the grass grow - you certainly are to be

[steve]

You have an amazing work ethic, It must be that you never have any free time to just read a book or watch the grass grow - you certainly are to be

I should really have been doing 'real' work, but I thought the problems merited a quick response. I'll just have to work longer to catch up.

[Synewave]

Hi Steve,

Firstly, many thanks for the massive amount of work you have put into the new builds. I'm now on 959 (I deleted the month.ini and year.ini beforehand as you suggested).

I've created a page for This Month and This Year Extremes using the new webtags here http://www.synewave.co.uk/gcsweather/thismonthyear.php

There seems to be a few strange records here, for example

Temperature & Humidity This Month This Year
Highest Temperature 15.7 °C at 10:43 on 02 November 15.7 °C at 11:12 on 02 November

I would have expected the times to be the same for the month and the year?

This is just an example of what I'm seeing, but there are other strange ones on that page.

[steve]

Highest Temperature 15.7 °C at 10:43 on 02 November 15.7 °C at 11:12 on 02 November

They're both saying the same time now. I suspect there are still problems with the initial setting up of the data from the logs. Hopefully, from now on they will be maintained correctly. I'm not sure I can be bothered to track down any more bugs in the initial setting up at the moment, I was in half a mind not to include that code in the first place. If you manually correct them (stop Cumulus and edit month.ini and/or year.ini) see if they stay reasonable from then on.

[steve]

You could also look in the diags log to see what errors were logged.

[Synewave]

Yes, I just looked again and most have corrected themselves, apart from Humidity.

Rain seems a bit odd, I'll see if I have some dodgy records.

Thanks Steve.