Sat, 21 Dec 1996 17:13:49 +0100
|
On Sat, 21 Dec 1996 16:03:26 GMT Daniel Norton <[log in to unmask]> said:
>Let me get this straight: LISTSERV works by validating the date field
>but ignores processing the time zone; SunOS inews validates the whole
>field and is broken?? GMAB.
LISTSERV validates the "Date:" field, allowing any time zone to be
specified, for the reasons that I have explained. Under no circumstances
will LISTSERV reject a message due to a bad "Date:" field. If the syntax
is mildly incorrect (no weekday, '96' instead of '1996', etc), it is
repaired, otherwise a new "Date:" field is generated.
SunOS validates the whole date and rejects messages with an invalid date,
thus losing valuable information which a human could have understood
perfectly well. I call this broken. I don't know if SunOS validates only
the numeric form or all forms, but if it also validates the three-letter
forms it is bound to throw away a lot of mail from non-US sources, which
I also call broken.
As I see it, RFC822 created a big mess with the time zone issue. The only
reasonable solution is to (1) encourage people to use the numeric form
(which has been done) and (2) categorize date fields as follows:
a. Fields which have a valid numeric time zone. The time zone can be used
for whatever processing may be desired.
b. Optionally, fields which have one of a number of well defined three
letter time zones (EST, PST, etc), for which no usage conflict is
known. These can be used as above.
c. Other date fields. The time zone information is meaningful only to a
human reader and the program should assume that no time zone was
specified.
Any program which ignores the factual existence of c is a program that I
call broken. On top of this, there is the fact that many mail readers
don't know their time zone. Pegasus for instance seems to always give you
GMT. That's one reason I personally never try to use time zones for much.
Eric
|
|
|