I had a question about these two fiddles (new as of 2022-01-13 12:15 UTC):

This [one](https://dbfiddle.uk/?rdbms=postgres_14&fiddle=6436ccb05a7a23328b1947534738e9d4) (PostgreSQL) and this [one](https://dbfiddle.uk/?rdbms=mysql_8.0&fiddle=b3d35ca16a389bf44d4e0a4b19f7703a) (MySQL). 

<TL;DR>

Following a comment by Jack Douglas, I performed the tests again with the fiddles above! I had the "anomaly" occur for every PostgreSQL (9.5 - 14) and MySQL (5.5 - 8)/MariaDB (10.3 - .6) variant available on db<>fiddle.

In desperation, I did a search and found (to my **great** surprise) that in GB from 1968, Sunday the 18th of February, 02:00 to 1971, Sunday the 31st of October, 03:00, there was [no change in the clocks](https://www.timeanddate.com/time/change/uk/london?year=1970) (use the dropdown to vary the decades), i.e. it was BST (UTC +1) - so, 1970-01-01 00:00:00 GB time was actually 1969-12-31 23:00:00 UTC - hence the problems!

So, it would appear that the servers are correct re. 1970-01-01 00:00:01 - however, I still think that the server should always be set on UTC to avoid confusion in this area. I certainly learned a lot from wading through all of this.

</TL;DR>



These problems don't occur on db-fiddle ([PostgreSQL](https://www.db-fiddle.com/f/2WjS8jhurEAtR5KyoVd3ik/0) and [MySQL](https://www.db-fiddle.com/f/2WKrvVe6uEEG6LobCUz92s/0)) nor on the (new) sqlize.online sites ([PostgreSQL](https://sqlize.online/sql/psql14/51c13207f150218fae15687c868a9bb2/) and [MySQL](https://sqlize.online/sql/mysql80/efd83317239f479df029feadaa5fab97/)). 

[Oracle](https://dbfiddle.uk/?rdbms=oracle_18&fiddle=098f2c709dcdfd3573f1cc706139a064) & [SQL Server](https://dbfiddle.uk/?rdbms=sqlserver_2016&fiddle=420b6cbc24ce6b912b7f98f3e2f7f943) don't seem to suffer the same issues on db<>fiddle


> Anyway, the point remains, that **unless** I'm missing something really obvious (wouldn't be the first time :-) ), there appears to be a problem somewhere (maybe between the keyboard and the chair)!


The **obvious** thing that I was missing is that it was summertime for 2.5 years from early 1968 to late 1971 in GB.


> My own (respectfully suggested) recommendation for db<>fiddle is that, since the service has now become a valued tool for the entire db community worldwide, the server(s) be kept on UTC at all times.



Despite the above correction, I **still** think that the servers should run on UTC and **not** GB time is to avoid issues like this which don't occur on db-fiddle.com or sqlize.online.




> Please don't hesitate to let me know if there are any queries/points/factors which I haven't made clear - I would be only to happy to rectify/clarify any/all of what I've written!

Still applies!