Personally, I always convert formatted time strings to a unix timestamp before doing time operations, and then covert the resulting time stamp back to a formatted time string if I need one.

Also, in all the databases I use on a regular basis, the dates and times are all stored as unix time stamps in the tables.

As an example, last week I wrote some code which estimated the load on the server and modified the DB queries based on the load and then calculated the estimated time to completion of that project in the log files, updated every minute. All the calculation were based on unix time stamp except the final log file entry, which I converted to my local time format for easy reading.

There are lot of people who will write code to do time operations on formatted time strings; but this is "a kludge" in my view, as time operations should be performed and stored as a unix time stamp and when a human wants to read the time, we then covert the time stamp to a local format based on the time zone of the user.

Also, let's say that you are in Brazil and your friend is in Japan. You want to do the same task at the same time (on a computer). It is best to specify the exact time as a unix time stamp, so you both will use the same time; and if you want to know the "formatted time" you can covert that time to a time string appropriate for your time zone.

Anyway.... that is my suggestion. That is what I always do...... all operations (processing) in unix timestamps.

YMMV