Re: [Pan-users] pan on windows 7

[Duncan]

Koos Jan Niesink posted on Sun, 19 Aug 2012 17:30:53 +0200 as excerpted:

> I downloaded the latest windows binary from Steve Davies' site
> (https://sites.google.com/site/paninstall/). Installation on windows 7
> starter went fine, the same goes for the configuration. However when I
> started to download headers from the newsgroups there is some unexpected
> behaviour. I can e.g. download the latest 300 headers. When I close the
> application or click on an other newsgroup some or all headers are lost
> from the 1st newsgroup. Have anyone else noticed this issue?

FWIW I'm on Linux (couldn't do MS if I wanted to as it's proprietary and 
I couldn't agree to various clauses in the EULA, including the one where 
they waive responsibility for the code... expecting me to assume 
responsibility for something I'm not even allowed to see what it does 
before I'm supposed to agree to take the responsibility for it!), but 
I've not seen that sort of behavior here.

However, there's a number of pan options that could appear to do that, if 
you're not aware of them and have them set incorrectly.

First, be sure they're not simply being marked as read, with the option 
to hide read posts set.  Check prefs, behavior tab, groups section, and 
ensure that pan isn't set to auto-mark-read when leaving group or getting 
new headers.  Then in the view menu, under header pane, make sure match 
only unread articles isn't set, or if it is, that it's indeed what you 
want.

FWIW, the way a lot of servers handle per-group message sequence numbers, 
having either of the auto-mark-read options in prefs/general/groups set, 
can risk missing posts.  What happens is that these servers do their 
numbering on an incoming server, then after spam filtering and the like, 
pass them on to the customer-facing front-ends.  But with the processing, 
sometimes the in theory sequential numbers get out of order, and with 
those options checked, pan likes to mark whole ranges as read, instead of 
only the specific messages that you /have/ read (or deliberately marked-
read manually), thus marking some read that were originally missing on 
the server, but that the server will suddenly "find", later on.  Keeping 
those options unchecked helps prevent that, so I always recommend that 
people uncheck them, even if they would otherwise prefer to set that 
option.

If you do like all remaining messages set as read when you leave a group 
(or before fetching more headers), use the select-all-headers option, 
then mark-as-read.  Because this only marks the selected messages read, 
it doesn't mark any read that the server hasn't delivered yet, so is a 
bit safer.  It's also worth noting that you can customize pan's hotkeys 
and make this a faster sequence to trigger via hotkey, if you do this 
regularly.

As for hiding/unhiding the already read messages, I definitely DO use 
THAT feature here, but I used the hotkey customization to set "r" to 
invoke that toggle, so I can toggle it between view-unread-only and view-
read-and-unread-together, with a single keystroke. =:^)  Additionally, 
it's worth noting that there's a toolbar icon for that function, that you 
can glance at to see what state it's in, if you need to.


Meanwhile, another feature that can make this appear to happen is the 
relatively new "actions" feature, configured on that tab in prefs, that 
can be set to do various things automatically, based on score.  Among the 
possibilities are auto-mark-read and auto-delete.  Plus, if you have auto-
download set in combination with the pref back on the behavior tab that 
auto-marks-read downloaded articles, and of course have pan set to hide 
already-read articles...


One more, tho this one's actually for full articles, not headers-only.  
Pan's default cache size is only 10 MB.  That's fine for many text-only 
users or for binary users that do direct download-and-save, but for 
people like me that prefer to set pan up to auto-cache (but not save), 
then come back after it has everything local so there's no waiting, 10 MB 
isn't very big!  I remember when I first started with pan and this caught 
me out!  I had setup the download and went to do something else as I had 
always done on my previous client (umm... OE, back in the day...), then 
came back expecting to have it all downloaded, or at least just be 
finishing up.  But it looked like it had only grabbed about a tenth of 
what I wanted!  So I tried again, and watched in horror as it started 
deleting the oldest few still cached articles just as fast as it 
downloaded the new ones!

Back then, disks were much smaller tho, and pan's max cache size as 
settable in the GUI was only a GB.  One of my first bug requests was to 
have that raised, to 4 GB at least.  Charles (the previous main dev) 
raised it to 20, just to be safe. =:^)  Then a few years later, he 
decided that setting was too complex for simple users, and removed the 
GUI setting for cache size entirely (tho it could still be set by 
directly editing the config file), so it was even HARDER for newbies to 
realize what was going on!  But now days, new dev, and the setting is 
back in the GUI.  And AFAIK, there's not artificial limit at all on cache 
size these days.  If you have a multi-tibibyte disk and want to use it 
all for cache, I believe it's possible, tho I actually haven't done 
binaries in years now, so I've not actually tested it lately.  (But I do 
have my text-only pan set to something like 5 gigs cache, with never-
expire set on the servers too, so I have several years of messages saved, 
but it's still actually less than a gig used so the 5 gig should last me 
awhile....)

FWIW, I also forgot that I had pan's view filters set to match only my 
articles at one point, and wondered why all of a sudden I was seeing no 
new messages!  (Of course, I had hide-read on too, and I'd read all my 
own messages already, so I couldn't see that only mine were showing 
up...)  I think I actually posted after a couple days, asking if the 
group was still working or something.  Then when /that/ post showed up, 
new and unread... it finally began to sink in that it might be on my 
end...  I had a rather embarrassing explanation to post after I 
discovered what I'd done! =:^/  But at least it wasn't the embarrassing 
binary posted to the wrong group mistake I've seen happen before.  So it 
could have been /much/ worse, I guess. =:^)

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman