On Mon, 22 May 2000, I wrote:
> Another alternative, is to avoid destroying the environment specifically
> in the -Meb case. (And any others which are relevant.) Maybe that makes
> more sense.
>
> Thanks for the report. I will do something about it in due course.
I have done a bit of thinking about this. On reflection, it seems to me
that flattening the environment was a bad idea, and should be removed.
You never know what settings might be required in some OS.
That leaves the problem of the timezones. I now have a new proposal:
. There is an Exim option called tz, which can be used to set a
timezone.
. The value of tz defaults to the setting of TZ when Exim is built.
. In any Exim run, if timestamps_utc is not set, Exim compares the
setting of TZ with the setting of tz.
. If they are different, it makes a copy of the environment, leaving out
any existing TZ setting, and adding (if necessary) the required TZ
setting.
. This copy is then used from then on.
It seems to me that this scheme will work in all operating systems. The
HANDS_OFF_ENVIRONMENT compiler macro can be abolished.
Any comments?
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
