[Pan-users] Re: feature ideas

[Duncan]

Steven Adeff posted <address@hidden>, excerpted below, 
on Sun, 14 Mar 2004 13:42:48 -0500:

> I'm gunna post this here and after some discussion make a feature req. bug
> report...
> 
> I recently discovered a neat program called nget. its a command line
> binary download app. It has a pretty arcane CLI but it has features that
> should be easy to implement into Pan. It has the ability to merge accounts
> to fill missing posts. This would be a neat feature for Pan to have. Even
> if it just called nget, having the interface as part of Pan would be nice.

I don't believe integration of a third party command line client is where
PAN is headed.

The general goal re PAN has always been something a bit like BNR2
(Borland Delphi/Kylex based "Binary News Reaper" for MSWormOS and Linux)
in terms of multi-server handling.  However, development in that
direction has always been put off pending conversion of PAN to use a
database library backend, probably SQLite.  That will make things much
easier, and anything done in the direction of multi-server now is
essentially wasted effort when the conversion DOES happen.  The problem
has always been that such a conversion isn't a small project in itself,
and the time had never seemed right.

However, the situation re a database backend is seeing new developments
as we write, as anyone following the pan devel list/group is aware by now.
PAN's "trunk" development is rather quiet right now, as Charles and Chris
have been busy at their day jobs and in meatspace, so now's a pretty good
time to work on the conversion, but for someone with decent database
experience volunteering to help out (since Charles and Chris HAVE been
busy).  As it happens, someone with just such experience recently DID step
out, and it's beginning to look as if the long awaited backend conversion
is about to happen.

If it does, it may take 2-3 months to hammer out the code and get
something working well enough for beta release.  Because this would be
quite a large change, it may be several more months and likely several
additional betas before things get relatively stable on the new backend
once again.  As with the 0.11-0.12 GTK+2 conversion, it's possible the
current stable implementation will continue with a number of additional
0.14.x releases, until the new implementation, probably to be released as
0.15.x, gets stable.  Thus, the new implementation would start out with
0.14.9x beta release numbers pre 0.15.0.  Assuming this is correct. as
mentioned, current 0.14.x may have a couple additional releases, but
no major feature additions, just bug fixes and etc, with 0.15.x then
taking the lead.

Post 0.15.0, the new backend will likely continue to stabilize thru the
rest of the 0.15.x series, and may not be as stable as 0.14.2 and the
future 0.14.3, until 0.16 or later.  However, once that backend is
implemented and fairly stable, additional features that have only been
talked about as in the future up until now, will be actually possible. 
Among these, truly integrated multi-server handling is likely to be near
the top of the list, as it too will be a major development, likely to
bring additional instability for a time, but once it stabilizes, likely
with 0.17 or so, additional functionality and features should have a fully
functional and stable base to build upon.

Another possibility would be that after the database backend is
implemented, versioning will jump to 0.50, indicating the major progress
toward 1.0, and the significance of the changes and requirements.  If that
happens, 0.60 may be the integrated multi-server series.  (Or it may just
jump to 0.20 for the database, and 0.30 or 0.50 with integrated
multi-server.)  In any event, integrated multi-server looks to be possible
in the 2005 time frame, early in the year or even late this year if
everything "just works" and there's lots of time for development, later in
the year if things don't work well and/or there are "meatspace" issues
keeping the devels busy.  Of course, all that assumes the transition to
the database backend goes ahead and is stable enough for  a beta release
by late Q2 or early Q3, and stable release Q3 or early Q4, this year.

As for nget, it wouldn't seem to fit into the plans (perhaps if gnet
hadn't already been implemented, but..).  If you are really interested in
something like that, not just in multi-server functionality for
PAN, perhaps starting a new nget graphical frontend project might be the
way to go.  But check out freshmeat and sourceforge first, as it's
entirely possible someone already has such a project started.

-- 
Duncan - List replies preferred.   No HTML msgs.
"They that can give up essential liberty to obtain a little
temporary safety, deserve neither liberty nor safety." --
Benjamin Franklin