Re: it seems like a bug

[The Wanderer]

Eric Blake wrote:

> According to Mike Limansky on 10/31/2006 12:16 PM:
>
> > Hello all,
> >
> > i use coreutils-5.94-r1 on gentoo linux. it's seems you have a bug
> > (possible in getdate.c).
> >
> > # date -d "2006-10-29 + 1 day"
> > Sun Oct 29 23:00:00 MSK 2006
>
> Not a bug.  This is due to daylight savings; the +1day adds 24 hours
> and the lack of an hour starts the computation at midnight, but Oct
> 29 is 25 hours long for your timezone of MSK, leaving you stuck in
> the same day at 11 pm.

The argument would be that not going forward 25 hours, rather than 24,
for the day which crosses daylight savings time boundaries in that
direction constitutes a bug - or at least a misbehaviour. (Naturally,
failing to go forward 23 hours rather than 24 when crossing in the other
direction would then be a similar bug.)

The counterargument would be that "+ 1 day" should always produce the
exact same time increment.

The counter-counterargument would be that since the actual
date-displaying code does treat daylight-savings-time days as being of
different lengths from others, the rest of the program should do the
same thing.

I'm not sure what the counters to that argument might be.

--
      The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.