[JB]

* MHonArc should understand most non-numeric timezones. May need
  some education. Logs show MHonArc is ignorant of AEST, MET;
  no errors reported under fukuzawa. See <TIMEZONE> resource.

[AL]
<TIMEZONE> does not list JST for Japan in the defaults - seems to be US
only.
But Japanese MUAs may avoid using JST and usually be set as numeric
since much other software also tends to be US timezone abbreviations
only. No error message suggests next most likely cause is misset
clocks/timezones, specifically no time zone at all (which might be taken
as GMT and produce approx 17 hour offset from Japan?).

Any light shed by checking the date fields of the wrongly sequenced
messages?

[JB]
* A single clock incorrectly set far in the future can do 
  very large damage. More than enough to offset the gains
  of Date: indexing.

[AL]
Not understood. Wouldn't it just result in messages from that 1 computer
going to the top of the index, becoming very noticeable as they pile up
there and resulting in email from users who notice that to the sender
suggesting a clock reset? -- Leaving aside that you have buried the
sender's address in HTML comments (discussed in separate message)

[JB]
* Agree, a multi-index solution may be a good way to go 
  for medium term. (If benefit outweighs cost)

[AL]
Computer resource cost should be small - just space for extra date
indexes and index pages which would be low in comparison to messages and
text indexes, with little extra processing. Main cost would be time it
takes someone who has already figured out MHonArc rcfiles to use the
facility for extra indexes ;-)

[JB]
* This is the type of issue the relational database
  folks love (like www.reference.com) where they can
  let the archive viewer decide stuff like this.

[AL]
Looks like a useful site, thanks for the pointer. (Not as elegantly
simple for adding a list:
http://www.reference.com/cgi-bin/pn/go.py?choice=mlistsuggestform
and does not seem to provide for adding old messages at all).

Reply via email to