On Wednesday 28 November 2012 16:18:30 Ralf Mardorf wrote:
> On Wed, 2012-11-28 at 10:40 -0800, Hilco Wijbenga wrote:
> > On 28 November 2012 09:04, Ralf Mardorf <email@example.com> wrote:
> > > On Wed, 2012-11-28 at 08:45 -0800, unruh wrote:
> > >> In linux.debian.user, you wrote:
> > >> > On Wed, 2012-11-28 at 08:44 -0300, Eike Lantzsch wrote:
> > >> >> Yep. Unfortunately Microsoft never learned in > 25 years that the
> > >> >> world has more time zones than they might have imagined in
> > >> >> DOS-times.
> > >> >
> > >> > They did and as I already explained, I want to have the local time
> > >> > for the BIOS too.
> > >>
> > >> Why? Linux is designed to run the system time on UTC and to always
> > >> interpret the time using /etc/localtime, usually into localtime. All
> > >> filestamps are raw in UTC but interpreted into localtime. It is just
> > >> silly to have the rtc/bios clock on localtime, and causes problems and
> > >> has absolutely no advantages.
> > >
> > > If I save BIOS settings as a file and the hwclock is set to UTC, the
> > > files don't get the German time. The BIOS is the BIOS, it's neither
> > > Windows, I don't use Windows, but nor the BIOS is Linux, so Linux can't
> > > "translate" UTC to local time, when I save BIOS settings.
> > So you spend most of your time in your BIOS? Then you save your
> > settings and compare that file to previous settings? And then you go
> > back and do it again? And again? Is that really how you spend your
> > time? :-) That seems unlikely. So why do you care?
> No and I won't explain why this could be important, snce it doesn't
> matter. But try to imagine that it might be important and that companies
> take care about this, among others, Microsoft does take care. And I
> still don't use Microsoft, but have other reasons to use local time.
> > > Under Linux I never noticed any disadvantage, when the hwclock is set
> > > to local time. Why should there be issues?
> > By and large, I don't think you will see any issues.
> Exactly, there are no issues when using Linux with the hardware clock
> using local time.
> I don't say that UTC always is an disadvantage, I just try to say, that
> for some usages it is an disadvantage.
> Nobody does explain for what usage UTC is an advantage.
> > Assuming you have
> > a proper time zone set, your computer will repeat an hour every year.
> > That might cause problems (if your computer is on during that hour).
> > But these problems are known so software might work around it. Lots of
> > people dual boot with Windows (which forces them to use local time) so
> > you should be okay.
> So still no issues when using local time, but still an advantage when
> using local time.
> > As others have mentioned: UNIX uses UTC.
> So until now no explanation why this should be better, but I already
> give an explanation, why this is less good.
> But somebody already blamed Microsoft as idiots. They are idiots, but
> not regarding to this issue.
That refers to me, but I didn't say "Idiots", I said "retards" and that is not because of their decision to use hwclock set to localtime but because they do ignore many timezones and do create problems for users in countries which do not appear in their list of timezones. I consider it retarded to ignore the needs of one's customers not only for years, but for decades
But maybe purchasing such goods is even more retarded.
Now back to Unix:.
> > That's the smart thing to do
Oops this statement appears now very much out of context but looks cute.
Take this as an example for HOW NOT TO CITE.
> > because it's reliable (independent of time zones and summer/winter
> > time) and every timestamp is fixed. By using local time you are
> > swimming against the current but I don't think the current is all that
> > strong. :-) It just seems to me that doing something that is
> > suboptimal for the sake of a BIOS' settings file's timestamp is a bit
> > silly.
> I don't have an disadvantage, I'm using Linux since November 2003,
> there's no Windows on my machine (excepted of XP on VBox). 9 years
> without an issue using local time, but the advantage that the BIOS can
> add the correct timestamp too.
You may do that as you please and if the timestamp on files saved from BIOS is so important to you it is certainly a valid reason to do so.
> So what is the advantage of using UTC? Until now I only see an
> disadvantage using UTC.
The advantage is that because UNIX is designed that way with worldwide networking in mind it is usually (but as you validly pointed out, not always) best to set up the system according to design criteria.
E.g. it becomes sometimes interesting when e-mails are exchanged with mail clients on different OS. Or think of banking transactions or stock exchange transactions.
Mind you that timestamps are not saved as date/time but as Unix time numbers according to the Unix epoch. The Unix time number is then interpreted by the system as date / time according to /etc/localtime.
Even under Unix there are different standards for doing this and of course there are also problems.
If there is a discrepancy between files created by BIOS and those created by the Unix system then it is obvious that the BIOS in question does not use Unix time numbers.
It could have been programmed that way or not. I wonder if there are BIOSes which use the Unix time number? SGI, SUN, experience anybody?
In the 1980s many of our customers (WANG) used local time on the hardware clock. That became an issue as soon as the systems started to be networked / interconnected with IBM SNA and later WANGOFFICE.
The problems which can occur if the hwclock is not set to UTC and the local time calculated according to timezone and the country's daylight saving arrangement is that files created during the shift from daylightsaving to normal time or back you might have diverse files with timestamps close to each other while in reality the files were created more than an hour apart.
This can be the case in the same way if you save files from BIOS during the timeshift, provided your hwclock is set by your system after you rebooted. Using ntp of course.
I admit that this problem might be purely academic as long as you have only one or a few systems to maintain but as soon as the systems are many and diverse as to operating systems, there are headaches preprogrammed.
Been there - done that.
> Errare humanum est, sed in errare perseverare diabolicum ;).
> If I'm mistaken and there should really be an disadvantage using local
> time, can somebody please explain it? If not, why cherish something that
> has the disadvantage I mentioned? And why always talking about
> Microsoft? This is a Linux list. The BIOS is needed by Linux too.
Because we aproached this from a more general standpoint and the case which you mentioned is a very special case. As I said: both standpoints are valid.
You state a case in which it is not helpful to set hwclock to UTC and we said that *in general* on Unix-like systems it is better to set hwclock to UTC.
Both statements do not exclude each other.
Yes, and best regards too