Yes, right from the start I used a custom schema, derived from wview, with floats rather than doubles. I figured that was more useful as I run a 1 minute interval.
I also trimmed out most of the unused columns. This did cause extra problems when version 5 came along as the default skin started recalculating some of the derived quantities and chewing up lots of cpu cycles. That gave me the choice of either adding the columns back in or tinkering with the skin. So I am happy with the result and have to remember not to complain when I get to reassemble the broken bits. And I know nothing about hail - except that the recent storm with 5cm hail took chunks out of the top of my rain collector and off the bottom of the Stephenson box. And one cup of the anemometer is missing (but I don't know it that was hail or cockatoos) On Tuesday, 18 November 2025 at 9:18:26 am UTC+10 John Smith wrote: > On Tue, 18 Nov 2025 at 04:06, Pablo Sanchez <[email protected]> > wrote: > >> One thought is to to ensure that the data types also cover metric. >> > > I store data as metricwx... > > In any case the tuning steps would be the same for imperial, check columns > for unique values, and set columns number types accordingly... > > However the biggest saving was changing the columns from double to float, > this would nearly halve things because you go from 8 bytes per value down > to 4 bytes, there is no need for using double because that allows for 16 > decimal places of precision and a much bigger numbers as well which is well > beyond what's needed when only 1 decimal place is usually required. > -- You received this message because you are subscribed to the Google Groups "weewx-user" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/weewx-user/ee0bd04d-bf06-4b3c-853a-32801f8e53f6n%40googlegroups.com.
