[Pan-users] Re: pcre error (was: undefined reference to iconv)

[Duncan]

"arnuld uttre" <address@hidden>
posted address@hidden,
excerpted below, on  Mon, 12 Jan 2009 10:17:41 +0530:

>> On Fri, Jan 9, 2009 at 10:33 PM, Duncan
>> <address@hidden> wrote:
> 
>> The recommendation is therefore to try the latest pan, 0.133.
> 
> okay, I thought PAN has 2 branches: stable and testing. I did not know
> that 0.14 was obsolete. Web-site did not offer much information on it.

Yes, that has been a bit of a frustration, both for me personally and for 
Charles, pan's primary developer.  

Charles had intended to have a stable new version, actually making it 
1.0, long before now.  Truth be told, with the C++ rewrite he restarted 
the versioning at 0.90 and I think they were supposed to be 1.0 betas and 
he didn't expect to ever hit triple digits before 1.0.  For over a year 
he released nearly weekly betas, rather burning himself out in the 
process, I guess.

But he obviously /waaayyy/ underestimated the amount of work necessary to 
get pan in shape for a real "worthy" 1.0.  The idea was that because the 
new C++ version was new code and modular, it would be MUCH easier to work 
with (and I think it was), and that all the features of the old code and 
more could be easily added back in.  Well, it's true... but there was 
simply vastly more work there than I guess he thought there was, and that 
along with sorting out some pretty tough bugs along the way, well...

So some 40 nearly weekly beta releases and a bit over a year later, I 
guess Charles decided, probably from shear burnout, that he had to take a 
break.  The last of those versions was 0.132, released on August 1, 
2007.  A year later to the day, he release 0.133 on August 1, 2008.  It 
included several patches that had been applied to 0.132 by various 
distributions and were thus submitted by the community.  One of these was 
a security patch, others included patches to fix pan's compiling with 
gcc-4.3+ and against glib 2.16+, and various i18n&l10n updates.  However, 
he's evidently not yet ready to get back into the grind, as there was 
little new code beyond community submitted patches, and as of a week or 
so ago, there had been nothing (beyond a handful of l10n updates, the 
work of GNOME's i18n team, not Charles) done in the subversion repository 
since.

Meanwhile, at least one major feature, the "rules" that allowed auto-mark-
read/delete/download, has yet to be ported to the rewrite.  According to 
discussions Charles has participated in, he doesn't intend to port it 
directly because the previous implementation was simply too complex to be 
practical for many users (it was, we helped quite a few users with it 
here over the years), but to reimplement the auto-actions, the bit that's 
actually practical in the new version, with the GUI as another tab in 
preferences, with options for each of the three auto-actions triggered 
based on the same score categories used on the color tab.

With that and some documentation, plus random bug fixes and maybe a way 
to categorize the subscribed groups into at least one additional level on 
the groups tree (the other most requested item, the workaround in old-
pan, multiple "servers" that could actually point to the same one, 
doesn't work in new-pan due to the full multi-server integration), pan 
would be pretty much ready for a 1.0.

However, that's likely to be another several months of coding and bug-
fixing all the bugs in the new code before it would be ready.  Charles 
has debate a 1.0 with basically what we have... but that hasn't happened 
either, maybe because he has decided my argument that to do so before the 
rewrite has at least something to replace the auto-actions possible using 
the rules in the old implementation, would be a big disappointment to 
those who have used pan in the past, and have waited so long for a 1.0.

But... Charles /had/ at least the intention of 1.0 being the next stable 
version, so here we sit.  The rewrite is by now certainly as stable as 
the old code ever was, scales VASTLY better than the old code ever did, 
and has some nice new features, but OTOH, still lacks at least that one, 
from the old code.

I've wished for years that some other developer with the necessary skills 
would take an interest and be able to work closely enough with Charles to 
take some of the pressure off, and hopefully get pan out of the fits and 
spurts cycle where it goes great guns for awhile, then all but stops, 
then goes great guns again, but it hasn't happened.  I wish I had the 
necessary skills.  I don't.  I don't know if it's that Charles is hard to 
work with, he hasn't seemed so from what I've worked with him, but as I 
said, I'm not a dev, and it'd be different if I was, or if there's simply 
not very many people, at least devs, interested in news any more, at 
least the GTK/GUI binary downloader text poster niche that pan fills.

I'm sure you can read between the lines some of the frustration, at 
myself for not having the skills, at those who do for not being 
interested, at Charles, tho I too would have been burned out after a year 
of weekly betas, etc, and can't really throw stones as I don't have the 
skills myself.

Meanwhile, pan remains in many ways the best solution out there, 
certainly for gtk (arguably klibido on the KDE side is a worth comparison 
as just a binary bulk downloader, and there are other clients for text), 
and while I can't code and help the project that way, I continue to 
volunteer here, helping at the level I can.  I've not tracked whether I'm 
the most senior active list regular or not, tho I expect I'm close, and I 
believe I am the most active, hopefully helping...

Not that all that's much help for your particular problems of the 
moment... but it should give you some background, for whatever /that's/ 
worth.

>> how old Centos 4.4 is, either.  If it's >4 years old, old-pan should
>> fit right in.
> 
> yes, CentOS 4.0 was released on March 2nd of 2005. The 4.x are just with
> new security updates keeping the base system same. Anyway, I tried
> installing 0.133 and got this error:


If it's that old and you've not updated them, then the gcc/glib/gtk are 
likely equally old, and old-pan may indeed be a better fit.  New-pan 
should work too tho as it's in theory backward compatible well beyond 
that.  Unfortunately theory isn't matching reality for you at this point 
and you're having problems getting either old or new pan to work.  
Frustrating!

> checking for PCRE... configure: error: Package requirements (libpcre
>    >= 5.0) were not met:
> 
> Package libpcre was not found in the pkg-config search path. [etc]
> 
> address@hidden pan-0.133]$ pcretest -s
> PCRE version 4.5 01-December-2003

OK, so you need libpcre 5.x and only have 4.x.  As I said, with something 
that old, maybe the old-pan would be easier, but either way, you're 
having problems and need to fix one or the other to work.

> If I install the newer pcre in /usr/local , will it break my existing
> system. I can't take the risk as its office computer, not mine. Or you
> guys can suggest some other older version or some other older
> Newsreader.

The problem I'm running into helping you directly is that I'm not a 
CentOS/RH/Fedora guy, tho I did run Mandrake, which was supposed to be 
Red Hat compatible, years ago. I switched away, to Gentoo, in 2004, and 
remain quite happy with it today.  But Gentoo works best for those who 
keep relatively up-to-date, which I do and that's one of the reasons it 
works so well for me, so I've long forgotten what versions of various 
things were current back in 2005.  So not only am I fighting "impedance 
mismatches" due to the distribution cultural differences, but I'm 
fighting the time gap as well and thus am finding I'm not as much help to 
you as I'd like to be.

FWIW, currently available Gentoo libpcre versions are 7.7 and 7.8... (I'm 
on unstable/beta/testing, which Gentoo calls ~arch, so am of course 
running 7.8) that's the gap we're trying to bridge.  So 5.x looks pretty 
ancient from here, yet it's newer than what you have.

What I was hoping to see were two or more different major versions, say 
5.x and 6.x, so I could check to see if they were slotted differently.  
Slotting allows installation of both, so it's a good hint that you 
/should/ be able to install 5.x without messing up 4.x.  But both 
available versions here are 7.x, and both the same slot, 3.x, which 
indicates that there has been parallel installation available in the 
past, but isn't now.  I've therefore no direct clue on whether 4.x and 
higher could be installed together without interfering with each other, 
or if indeed, 5.x might actually be backward compatible with 4.x.

FWIW, meanwhile, there's a Fedora regular here too, who contributes where 
he can, but he's not particularly technically inclined, so that's not 
going to help much either.

So we gotta go a different way...

Let's backup a bit.  You haven't mentioned knowing about pan's overall 
requirements, so I'm guessing you haven't seen pan's requirement page, 
here:

http://pan.rebelbase.com/requirements/

Taking a look at that list, is the pcre 5+ requirement all you're lacking 
on the 0.90+ requirements?  If you are lacking several of them (and 
they're not easily available from the CentOS or equivalent repositories), 
it's probably easiest to shoot for 0.14.x, even if it is not any longer 
officially supported.  If pcre's all you're lacking, fixing that and 
being able to run a current pan will likely be better, and may well be 
easier.

Next question, CentOS 4.4, what does that correspond to for Red Hat 
version, and if you know, Fedora version?  Perhaps we can find you some 
compatible repositories and pan and dependencies can be filled from 
binaries from them.  While we're at it, what repositories do you have 
configured, currently?

As for news client alternatives, there's lots of them out there, but few 
filling the generally well rounded niche pan does anywhere near as well.  
But if you're not using much of what pan offers anyway, there may well be 
a client either more lightweight, or better suited to the specific 
purpose you had in mind.  I had avoided asking earlier as you hadn't 
volunteered it and I had no need to know, but to answer this one, I do, 
what sort of groups are you intending to use your reader, once you get 
one, with?  Only binary groups?  Only text/discussion groups?  A mix?  If 
binary do you know if yEnc posting is common in those groups or not?  
(Many clients don't do yEnc.)  Are you mainly interested in a binary 
harvester only or do you plan on posting, and if so, text posts only, 
binary only, or a mixture?  If binary, mostly single-part such as jpeg 
groups, mostly multi-part such as ISO, movie, and to a lessor extent MP3 
groups, or both?  If you do both text and binary, would a separate client 
for each be a big bother or do you strongly prefer a single well-rounded 
client such as pan?  Are you most comfortable with a GUI or would you 
find a terminal mode client usable?  Do you do EMACS?  (If so, you may 
find gnus a good fit.)  Do you have GNOME installed?  GTK (presumably yes 
on that since pan uses it)?  KDE?  Are they available and do you mind 
installing them if they aren't installed already?  Which versions 
(major.minor)?

Obviously, answering these questions will help in recommending suitable 
pan 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