time_t - When does it break?

[Guillaume Lessard]

When does it break?
Every time the computer's time zone is changed, the time stamps of  
every transaction in the file get updated relative to the new time  
zone. This is dumb. This is broken.

In what world does traveling change the past? Not this one.
With gnucash, traveling does change the past. Great!

Yet, when I reported this, it was closed as "not a bug."
Oh, was it *designed* to misbehave?

Clearly, this project has serious time-related issues...


What I get from the discussion:
- time-of-day is useful for transaction ordering, and not much more.
- financial transactions occur on a given date, regardless of the actual
   moment within that day.
   (Interest calculations and tax rules are done in terms of dates,  
not timestamps.)
   (The rules under which a given transaction occurs won't ever change  
to the point of
    requiring a change of date. Especially not retroactively.)

What information is needed in a transaction:
- the day it occurs, for financial purposes (no time stamp.)
- the time at which it was captured and/or last modified (this is a  
time stamp)
- some want to be able to order things, this may be easiest done with  
separate time-of-day information.

As it is, the information stored is a time stamp that does *not*  
include time-of-day information, and changing the bogus time-of-day  
portion of that time stamp will *not* fix the problem. No matter how  
much eye-rolling anyone does. The Earth isn't flat.
(I acknowledge, however, that it would be possible to further kludge  
the use of that time stamp so as to make it work. Documentation  
nightmare.)

If anyone has different conclusions, put them all in one message, so  
that the discussion can work...

--
Guillaume Lessard
glessard at tffenterprises.com