Re: [Exim] Those Darn Timezones - experimental evidence need…

Top Page
Delete this message
Reply to this message
Author: Philip Hazel
Date:  
To: exim-users
Subject: Re: [Exim] Those Darn Timezones - experimental evidence needed
Thanks to everyone who sent me data. The majority of systems seem to
revert to wall clock time if there is no TZ variable in the environment.
This is clearly documented for Linux, which claims it conforms to SVID
3, POSIX, and BSD 4.3. The systems that behave differently are, ... wait
for it ... surprise, surprise ...

  AIX      which gives you GMT
  HP-UX    which gives you EST (yes, indeed!)
  IRIX     which gives you GMT
  SCO_SV   which gives you GMT (documented as "implementation defined")  


I'm not sure about DGUX, because the only test was run in the GMT zone.
(Ken, can you try again once we are on BST, please?)

What I have done is to implement conditional code to remove the entire
environment (since Exim makes no use of it, and flattening it all is
easy), and set up the four OS above to opt out of it. If anybody can
tell me how to enforce wall clock time in those OS, I'll consider adding
something special for them.

I currently have no information about the following systems, but have
not opted them out.

  GNU/Hurd     likely to be same as Linux since it uses glibc
  HI-OSF
  HI_UX
  mips 
  QNX
  ULTRIX
  UNIX_SV
  USG
  Unixware7       


I will also implement a "do all logging in UTC" option, but the default
will not change.

-- 
Philip Hazel            University of Cambridge Computing Service,
ph10@???      Cambridge, England. Phone: +44 1223 334714.