[Pan-users] Re: Re: Pan version number from CVS sources?

[Duncan]

walt posted <address@hidden>, excerpted below,  on Fri, 28 Apr
2006 18:33:31 -0700:

> Well, maybe you can answer one more puzzle for me?  I compiled 0.94
> from the source tarball, but it just is not a complete version of the
> pan that I know and love  :o/
> 
> For example, 0.14.20 has eleven menus on the top menu toolbar;
> 0.14.90 removed the 'Tools' menu and redistributed those items
> amongst the other ten menus.
> 
> BUT -- 0.94 has only six menu items on the top menu bar:  the
> 'Server' menu, for example, is gone so I can't switch between the
> news servers I defined with the 'Add Server' dialog.

OK, you apparently missed the announcement of the new version, and are
understandably confused.

What happened is this:

Charles went on what we all thought was an extended vacation, doing very
little with the 0.14.x pan that we all
know/love/hate/at-least-are-familiar-with.  That would have been fine, as
pan is software libre, Charles is volunteering his own time to write it,
and as a volunteer (a few tens of dollars in the tip jar is all... over
how many years now???), if real life leaves  no room for everything...
well it's understandable if that volunteer coding time gets shoved aside.

That said, when the shove-aside seems to be two years... you begin to
wonder if the project is dead...

Surprise, surprise!  It /wasn't/ such an extended vacation after all, but
instead, an entire rewrite into C++, now more modular, using far, far less
memory in large groups.   The several million headers/overviews in some
groups on some servers wouldn't even load before, even with multiple gigs
of ram and left to churn for an hour!   Now it loads them in half a gig,
typically in less than a minute.

The biggest additional feature is the automated multiserver handling. 
There's a reason there's no server menu now!  It's not needed, because PAN
combines everything into a single view.  The significance of this is that
where the same group is carried on multiple servers, just as pan used to
automatically manage multiple connections to a single server, pan now
manages multiple servers (with multiple connections to each).  Instead of
separately setting up download jobs for each server, you set up a single
combined task list, and pan manages it automatically, feeding individual
tasks to only the server that carries that group if only one carries it,
feeding parts out individually to each connection to each server, if
multiple servers carry the group, just like it used to handle multiple
connections to the same server.

The basic code is there and generally working, and improving with each
week's release (five releases now, 0.90-0.94 in five weeks, one a week,
typically about Sunday).  However, there are still some rough edges and
certain bugs.  It's no where near as slick and smooth as 0.14.x was after
many years, but it's improving faster than pan ever has before, because
the code is smaller and more modular and easier to work changes into,
without them feeling like (and sometimes being) hacks that have a good
chance of introducing bugs elsewhere due to the nature of the code.

As for some of the other missing features, that was deliberate.  The idea
was to see what folks complained about being missing, and return it only
as necessary, dropping functionality and options that nobody was using, or
valued enough to trouble themselves to request again in the rewrite.

Since it's the weekend already and 0.95 will likely be out in a day or
two, I'd suggest running 0.94 enough to get the general idea, but don't
get too worried about things or make requests just yet.  When 0.95 comes
out, compile, install, and try it.  If your issues haven't been
resolved yet, /then/ post a list of what you need from the old version
that's still missing, plus any bugs you are experiencing, preferably by
Tuesday, and there's a decent chance at least a couple of them will be
resolved with the next weekly update, 0.97.  Some may of course take a bit
longer.

Meanwhile, if you find it's still lacking critical stability or
functionality, continue to use  the old 0.14.x version for most stuff. 
I'm still doing that at present for text groups, and using klibido when I
do binaries.  (klibido likewise does multiple servers, but is a bit more
polished and less buggy than 0.9x pan has been so far.  I expect I'll be
switching back to pan for binaries by 0.100-ish, tho, six weeks, possibly
a couple months if Charles takes a few days off and skips a weekly release
or two in the mean time, as pan is really evolving that fast, at this
point.)

The biggest thing holding me back on 0.9x now is a (previously posted) bug
that's causing an assertion fail and a failure to display certain post
bodies -- enough that it's not workable for me yet.  That may be amd64
only or may be related to the posts in question.  I'm not sure, and I'm
not a C coder so am not as much help as I'd like to be tracing and
debugging.  Before that, the biggest problem I had was that I have an
apparently unusual color scheme -- light text on a  dark background, and I
couldn't read some of the default-colored content.  Charles had removed
those color options from the prefs when he removed everything else, but
after I mentioned that it made PAN unworkable for me because I couldn't
read some things due to hard-coded colors clashing with scheme colors,
Charles put the option back for changing the various quote colors and the
like, and I could again actually read the content.  That made it workable
enough to find just how bad the first bug mentioned was, so that's what
I'm waiting on now.  A couple more cycles, and the new pan will be
the best news solution I've got, as old pan was for many years and still
is for text and the occasional binary.

-- 
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 in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html