Ubuntu affecting Windows XP clock

[John]

Steve McIntyre wrote:
> On Fri, Sep 24, 2004 at 01:23:31PM +0800, John wrote:
> 
>>Kevin Krumwiede wrote:
>>
>>>The hardware clock should always be set to UTC.  It just makes things 
>>>simpler, especially for DST changes.  Unfortunately, this is another 
>>>example of Windows being fundamentally broken and everyone else having to 
>>>work around it.
>>
>>How does it make things simpler? You've offered no argument to support 
>>your position.
> 
> 
> If you store local time in the RTC, there is no way for different OSes
> to know if any DST changes have been applied. This leads to ridiculous
> situations (like the first Win95 release) where an OS / several OSes
> may each try to put the clock back/forward when they're booted after a
> change.

There's not?
The tz environment variable that can be set on OS/2 as I recall (and 
it's a long time since I used OS/2, longer ago since I had to cope with 
DST locally) is that it has the transition dates.

I would expect the difference from standard time might be applied to the 
code that interprets the hardware clock.

What actually happens I don't know, but it's wrong to say that the 
hardware clock _has_ to be changed.

> 
> The only sane way is to store UTC in the RTC and apply offsets when
> you read/write to it. Then everybody's happy. The only reason Windows
> still does the broken-by-design localtime thing is for backwards
> compatibility with DOS, which didn't know any better.

Indeed, early PCs didn't have a non-volatile clock. I think that came in 
with the AT.