[Pan-users] Re: Multiple server support even working?

[Duncan]

Phil <address@hidden> posted
address@hidden, excerpted below, on  Thu, 03
Apr 2008 03:12:01 -0700:

> The old Pan DID do this, like any reasonable newsreader should.  But in
> their wisdom the developers actually *removed* this expected feature in
> the new C++ Pan. So if you want it, "upgrade" to old Pan.

That's not really correct.  The new/C++ pan is a rewrite from scratch, so 
no feature has been removed, it simply might not have been added to the 
rewrite yet.

Additionally, an anti-bloat strategy Charles took was that any "extra" 
features in old-pan weren't added to the rewrite unless/until they were 
requested.  If nobody's using them, save the code and complexity, and let 
the previous feature RIP.

For over a year there were nearly weekly betas.  Those who tested and got 
their requests in early generally got their features.  Those who 
didn't... well... didn't.

> I posted some time ago requesting this feature be put back.  Looks like
> it hasn't been.

Some time ago?  After (as I said) over a year of nearly weekly betas, 
Charles took a well deserved break.  That has been about 9 months now, 
IIRC, and during the last six months or so of the weeklies, he wasn't for 
the most part actively implementing new features but rather, beating the 
bugs out of the existing code.

So how long ago /was/ it that you asked, and how active were you in 
finding others that wanted/needed the feature as well?  Just one request 
is easy to pass off; many, not so easy.

> I hate to whine about any useful open source project,
> but the field of available linux newsreaders is quite small and on this
> one development is certainly slow ....

> Is there a need to add developers to the team, presumably fluent in C++?
>  If so, has there been a call for volunteers?

I'd say there's some need, yes.  Charles is great when he's active, but 
he tends to go in spurts.  The weekly betas were great, but that was 
after over two years when to all outside appearances the pan project was 
dead, as he did the rewrite privately, nobody seeming to know what was 
going on.  Now, it has been another nine-ish months (I believe, I've not 
been counting, exactly) of little externally visible (even in SVN, which 
I run) activity again.  I'm sure he was burned out, but...

So yes, some good C++ coder help would probably be a good thing, and 
could help to stabilize development. 

I honestly can't say how easy Charles is to work with at that level, or 
whether indeed he'd let someone take the lead during his "off" pulses or 
in general.  I do expect that anyone wishing to work cooperatively would 
need a strong appreciation for some of the current pan goals and 
development rules -- in particular, Charles has spent a LOT of time and 
effort on getting and keeping pan's 100% GNKSA certification, and if 
anyone has proposals that might change that, he's made it quite clear 
it'll take a hostile fork to do it.  I believe all the project and list 
regulars support that stance (I absolutely do), or they'd not be regulars.

I believe the problem has been that to most of the world, including most 
C (old pan) and C++ (new pan) developers, apparently, NNTP is pretty much 
a dead protocol.  For some reason, that seems even more so in the FLOSS 
world, maybe partly due to the fact that many of the closed source NNTP 
client projects are either pay software or support specific pay sites and 
their featureset (as was the initial nzb story, if I'm not mistaken, tho 
it's open enough it has become more or less a working standard now, if 
not an official one); IOW, it seems there's simply not enough interest to 
create a viable FLOSS NNTP community.  

Actually, that applies to servers as well altho there may be 
justification there for financial sponsorship, as it seems NNTP server 
software development is pretty stale too, with no REAL current FLOSS 
community pressure on High Winds, the main proprietary/commercial player 
in the NNTP server market.  And they could really USE some competition, 
as their product certainly has the performance when it's tuned and 
working in peak form, but it seems to be extremely difficult to keep 
tuned and working in that peak form, as users of servers running it can 
well attest.

Really, the obscureness of NNTP remains both a curse and a blessing.  For 
binary users, the obscureness and somewhat the inefficiency of the 
protocol and relative difficulty of use of the clients, has been what has 
kept it off the radar of the various censors, be they copyright cops or 
political or moral censors.  Were it ever to reach the popularity of say 
the various P2P apps, it'd be heavily targeted, and would have likely 
gone the way of Napster and the other more centralized server systems, 
for much the same reasons.  However, that very obscurity has kept it 
amazingly useful for those that know about it and have the patience and 
skills to work the system.  The thing is, however, that there's a dynamic 
balance.  Every time it starts getting attention, the censors and 
spammers come in, dropping the attraction once again.  That drops it into 
relative obscurity once again as the cost for payoff ratio for the 
ordinary joe gets too high.  Of course, that reduces the relative 
interest of the censor forces as well, who find bigger fish to fry, 
leaving it in peace long enough to become slightly more effective and 
draw more attention once again, thus repeating the cycle.

So the obscureness and with it the relative scarcity of development 
resources is both a blessing and a curse.  To be sure, if we suddenly got 
all the development and user attention we wanted, we'd soon find the 
protocol not worth our time.  But that will never happen because the 
balance will be dynamically forced back toward relative obscurity (and 
thus usefulness only for the few who know) before it ever reaches that 
point.

But that's a digression from the topic...  Anyway...

Yes, FLOSS could really use some attention in the area of NNTP clients, 
and certainly pan as such a client is no exception.  If you are a 
developer or have developer resources at your disposal and are 
interested, I'd certainly urge you to contact Charles and talk things 
over.  Whether it'd come to a fork or not, I really don't know, but 
either way, pan could certainly use more developer resources devoted in a 
steadier fashion than Charles has proven able to give, alone (or even 
with Chris' help, tho I don't know how active he has been since the 
rewrite).

Meanwhile, there are a few other alternatives.  klibido is the big one 
for binary harvester (or bnr2/3 if you don't mind open source based on a 
closed and now deadended on Linux compiler system, Borland's Kylex/
Delphi), and there are a decent number of text client alternatives.

-- 
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