On Thu, Aug 19, 1999 at 04:18:17PM +0200, Thomas Roessler wrote:
> On 1999-08-19 09:29:03 -0400, Fairlight wrote:
>
> > Now....I could see it saying that /var/spool/mail/fairlite had no
> > new mail...maybe irc does something when it checks my mailbox.
> > BUT... -Z should -not- tell me there's no new mail in any of my
> > mailboxes when I have two freshly generated mailboxes with a new
> > message each in them, should it?
>
> This is a known limitation of the way how mutt's new mail detection
> algorithm works. I'd suggest you set save_empty for your incoming
> mail folders.
I'll probably do that, now that -Z will easily let me check for new mail.
Thanks for clarifying the matter.
If I could be so bold, I -think- I have an algorithm that, if implimented,
would fix the limitations:
.muttncrc - Mutt New Check rc
mailboxname datestamp size
mailboxname datestamp size
...etc
algorithm -
1) If new mailbox determined from .muttrc, add to .muttncrc, current
timestamp.
if size > 0 bytes, new mail in mailbox, set size in file
if 0 bytes or nonexistent, set size 0 in file
2) If entry in .muttncrc already, check timestamp against mailbox
if same, no new mail
if newer, procede to size check
if larger, new mail present
if smaller, no new mail present
(both cases), set new size and timestamp
3) Retain current during-run algorithm
ideally, your algorithm works fine, and timestamps/sizes should be
set during deletion and arrival phases, because relying strictly on
startup-state data could potentially be inaccurate. this could be
done in memory potentially to minimize overhead, although I've
noticed mutt not complain about multiple instances running on the
same folder, so theoretically either a lock condition should persist
or the timestamp/size should be written at each change in condition.
The current algorithm in place obviously relies strictly on data from point of
binary start. What appears to be needed to get around these limitations is
the equivalent of a .newsrc, but for time and size, to maintain states
between runs.
I -think- this would work. I realize this is probably best on the
developers' list, but I'm not really up to becoming a real developer on
it...it's just an idea I'm tossing out.
Even with the limitation if it's not addressed, it still kicks elm's arse.
:) Thanks!
mark->
--
Fairlight-> ||| [EMAIL PROTECTED] | Fairlight Consulting
__/\__ ||| "I'm talking for free... | http://www.fairlite.com
<__<>__> ||| It's a New Religion..." | [EMAIL PROTECTED]
\/ ||| PGP Public Key available via finger @iglou, or Key servers
