[Pan-users] Re: pan servers settings

[Duncan]

Jethro Tull <address@hidden> posted
address@hidden, excerpted below, on  Wed, 27
Aug 2008 22:13:56 +0000:

> Actually I'm using hotmail. this enables RTF by default. now I disabled
> it, it should be ok.

Yes. =8^)

>> Second, you don't mention what version of pan you are using,
> since i posted for another issue just before in which i mentioned the
> version number here i completely forgot.

I hadn't gotten to that one yet (I had newer posts sorted first).

>> Newer pan checks all configured servers that have a particular group,
>> seeing what posts are on all of them.
> you mean pan uses all available servers to update news? for a given
> group, say alt.os.linux, all the posts should be same either on one
> server or the other, is it? maybe except that some servers are late to
> update. so why does pan uses many servers?
> anyway if Pan finds one post same as what it found on another server
> will it download it anyway?

Yes, pan uses all servers.  However, pan tracks by Message-id which is 
the same across all servers, and only downloads it in one place (unless 
you tell it to download it again after it has expired from cache or 
something).  In fact, it does the same for cross-posts too, downloading 
it once if you read both groups and marking it read in both.

The way it works is that pan starts downloading one post from the first 
server in a rank (yes, that's what I meant by priority), starts another 
on a second connection... up to the number of connections you allow, and 
does the same for other servers in that rank.  Thus, if you have four 
connections each to two servers, it can be downloading eight separate 
parts at once.

If one server has less parts than the other, pan will skip ahead, 
downloading all the parts it can get from it until it has everything on 
that server.  Since that server has less parts, after the first few, pan 
will be ahead on that server anyway, so it'll naturally get all the parts 
it can from the server with the least parts, downloading any fills it can 
from those with more parts as it gets to them on that server.

Once it has finished downloading what's available on all servers at a 
specified rank, it goes on to the next rank.  (Note that while the GUI 
only shows primary and backup ranks, two ranks total, by editing 
servers.xml directly, you can create as many ranks as you have servers, 
if desired of course.  If your server allows more than four connections, 
you can do the same there.  Pan has a number of such "advanced" settings 
only available by editing its config files directly.)

> you mean "server's rank" in Pan "servers settings"? i have already tryed
> it.
> actually, wherever i am only one server works. Whatever the priority
> settings i always get the error message that one server is off and no
> news are downloaded at all.

Yes.  That's what I explained later.  Pan checks all configured servers 
for messages, and will complain if it can't get ahold of one.  
Unfortunately, there's no way to tell it to continue anyway, or to 
temporarily disable certain servers, thus the workaround I suggested.  

(When we were first playing around with this feature, we tried setting 
the connections for a server to zero, hoping pan would interpret that as 
a disabled server.  Unfortunately, it didn't work.  It still complained 
and refused to go further, now without even trying to connect.  Still, 
some of us hope that Charles will get around to fixing it up so it works 
that way at some point.  Of course, if someone that new how to code  
provided a patch for it, adding say a "disabled" option in the GUI and 
making that 0 connections in servers.xml, it'd probably get in faster.)

>> The workaround, if you don't want it to check all your servers (say you
>> can't access some at all part of the time), is to configure multiple
>> server sets (using multiple servers.xml files and switching them out
>> manually).
>>
> ok i'll try doing it.

>> If you need help with the shell scripts, just say so, and I can post
>> samples you can modify to fit your needs.
>>
> I think:
> ##-------
> cp ~/server1.xml ~/pan2/servers.xml
> pan
> ##-------END
> 
> is enough?!!

That should work, providing you keep the source server files in ~, of 
course.  I was thinking more along the lines of putting them all in the 
same dir (~/pan2 by default), and just switching them out in place.

> but will kde run it again from the same script to run pan at next startx
> if in previous session there was a pan session running at logging out?

If you have it setup to do so, it should, yes.  Tho it really shouldn't 
matter since once it's copied over, pan would continue to use the same 
one every time until the other one is copied into its place instead.

What I did here is setup something like this, for several different
pan configs, PANDIR is set to a base dir (~/pan/) in my login scripts:

#!/bin/bash
# set the gtk2rc files path, note the wrap
export GTK2_RC_FILES="/etc/gtk-2.0/gtkrc:~/.gtkrc-2.0:~/kde3.5/share/
config/gtkrc-2.0"
# set pan's config dir
export PAN_HOME=$PANDIR/bin
# set a working dir, at one point pan was dumping misc files
# so I set this up so it wouldn't dump them directly in ~
cd ~/pan/scraps
# execute the pan binary, passing any command line options
exec /usr/bin/pan $*

I have three different scripts, each the same except for the line
setting/exporting PAN_HOME, which points to the bin subdir for my binary 
instance (above), the text subdir for my text instance, and the test 
subdir for my test instance.  Simple enough. =8^)

You could do similar, but probably don't need the scraps thing any more 
and as long as you don't need multiple instances, you'd delete that, but 
add the lines shuffling the servers.xml files.  Of course, you'd set your 
the gtk2rc files path appropriately as well, if necessary:

#!/bin/bash
# set the gtk2rc files path, note/correct the wrap
export GTK2_RC_FILES="/etc/gtk-2.0/gtkrc:~/.gtkrc-2.0:~/kde3.5/share/
config/gtkrc-2.0"
# switch the servers config
cp ~/.pan2/servers1.xml ~/.pan2/servers.xml
# execute the pan binary, passing any command line options
exec /usr/bin/pan $*

You'd then call that one pan.serv1 (or whatever) and setup another called 
pan.serv2 (or whatever).

Here, I changed the original pan kmenu entry to point to one of them 
(text), then added another pointing at another (bin), and only run the 
third once in awhile so run it from the open dialog only.  You'd point 
yours at pan.serv1 and pan.serv2, or whatever you called them.

If you do it this way and have KDE set to relaunch stuff opened when you 
last shutdown, presuming you had pan open, it should launch the same 
kmenu entry as you had run.  However, as I said, since you're just 
copying the files, you could in fact forget about launching pan from the 
scripts at all, and just have them copy the appropriate file over, then 
start pan normally.  You'd then only ever run the copy scripts when you 
wanted to switch from one to the other.

Here, I setup khotkeys entries to launch the the two, as well.  Actually, 
that's the only reason I made menu entries for them, was so I could setup 
the hotkey launchers, since I basically never use the kmenu for anything 
I regularly run, using a hotkey launcher instead, and only use the kmenu 
for apps I don't run regularly enough to have a hotkey for.  Of course, 
this bit is entirely optional.

-- 
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