bug in date washing

bug in date washing

[Paul Stevenson]

In article mode 'W T i' turns the date line into iso8601 format. Trying
to rewash the date into any other format then barfs.

My lisp skills are insufficient to provide a fix, but I have traced the
problem to the fact that article-date-ut uses the washed date line from
the header as input and passes this straight on to
article-make-date-line which itself calls date-to-time which calls
parse-time-string which does not understand iso8601 format. So either
article-date-ut could get the 'unwashed' date and pass it on, or
parse-time-string could be persuaded to speak iso8601.

Paul

Re: bug in date washing

[Kai Großjohann]

Paul Stevenson <spaul@mail.phy.ornl.gov> writes:

> In article mode 'W T i' turns the date line into iso8601 format. Trying
> to rewash the date into any other format then barfs.

On your message, I get the following:

W T i           produces correct ISO date
W T l           produces Date: Thu Jan  1 01:00:00 1970 +0200
W T l           produces Date: Thu Jan  1 00:00:00 1970 +0200
W T l           produces Date: Wed Dec 31 23:00:00 1969 +0200

I'm pretty sure that repeated W T l invocations are not supposed to
change the date, are they?

kai
-- 
Life is hard and then you die.

Re: bug in date washing

[Jack Vinson]

And when I play with Kai's message, I get the following:

Originally date header: Date: 19 Jul 1999 13:00:44 +0200
W T i           Date: 19990719T130044
W T i (again)   Date: 19691231T180000
W T l           Date: Wed Dec 31 18:00:00 1969 --500
      (clearly incorrect)
W T o           Date: Wed Dec 31 18:00:00 1969 --500
      (shouldn't this go back to the original?)

The only way to get back to the original date is to t'oggle the headers
twice.  Many of the other gnus-summary-wash-time-map bindings don't work
very well either.  If I change to elapsed time, I get the X-Sent header,
but it never changes to anything else after that.


-- 
Jack Vinson <jvinson@unagi.cis.upenn.edu>
Zippy: I have no actual hairline...

Re: bug in date washing

[Paul Stevenson]

Jack Vinson <vinson@unagi.cis.upenn.edu> writes:


> The only way to get back to the original date is to t'oggle the headers
> twice. 

Which probably is also a bug, since toggling the headers ought not to
re-wash any of the headers (IMHO).  Alternatively 'g' fixes things,
too.

Re: bug in date washing

[Lars Magne Ingebrigtsen]

Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:

> I'm pretty sure that repeated W T l invocations are not supposed to
> change the date, are they?

Eurm.  Well, uhm, probably not.  But then the original date should be
stored somewhere...  (This used to work OK before, because then we
just looked at the article header structure, but we can't do that now, 
since the function may be called on a multipart that contains a
message/rfc822 part or the like.)

Local variables won't work...  Text props?  Stuff the original date in 
a text prop, and then use that?  Yup.  Fix in Pterodactyl Gnus v0.97.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen