Re: [Pan-users] Re: Pan error part_mid_buf_len' failed.

[Keith Richie]

On Thu, Mar 20, 2008 at 8:41 PM, walt <address@hidden> wrote:
> On Sat, 15 Mar 2008 09:49:04 +0000, Keith Richie wrote:
>
>
> > Wondering what causes the error below
>  >
>  >
>  > Queued up a few files to download. Had to do some other stuff, so I
>  > closed Pan. After a while, Pan wouldn't start, and the error below was
>  > spout out on the command line -
>  >
>  > pan: parts.cc:244: void pan::Parts::set_parts(const pan::PartBatch&):
>  > Assertion `pch == part_mid_buf + part_mid_buf_len' failed.
>  >
>  > I'm using 0.132 from CVS with the image and nzb patch included.
>  >
>  > I managed to "fix" it by deleting the tasks.nzb file from .pan2/. It
>  > happened to be 7.5mb large. Wondering if it was a corrupt post I had
>  > queued up? Or if there's a size limit to the tasks.nzb file? I did open
>  > the file in nano, but it became quite tiresome searching for anything
>  > erroneous in the huge amount of text. The first task, was just fine.
>
>  The tasks file you gave us certainly makes it easy to reproduce the
>  problem :o)  The assertion is triggered so early in the program that
>  the bug *must* be blatantly obvious to someone like Charles (or maybe
>  someone else out there?).  Unfortunately Charles is busy making a
>  living (let's hope) and so meanwhile we debug the best we can.
>
>  The gdb output I get is:
>
>  Breakpoint 1, main (argc=0, argv=Cannot access memory at address 0x4)
>  at pan.cc:280
>
>  That error is so absurd that even I should be able to track it down
>  when I have some time.
>
>  Questions for you:
>
>  I think you and I use pan quite differently, so please explain exactly
>  what you do to queue up files to download -- e.g. are you offline when
>  you select the files, or what?  Why doesn't pan start fetching them
>  immediately?
>
>  Was pan already downloading files when you quit?
>
>  The 'tasks.nzb' is generated by pan, but in the case of the particular
>  tasks file you posted,  did you point pan at third-party nzb file(s)
>  for pan to import, or did you select articles manually?
>
>
>
>
>
>
>  _______________________________________________
>  Pan-users mailing list
>  address@hidden
>  http://lists.nongnu.org/mailman/listinfo/pan-users
>


Good questions!!

The tasks are from headers inside pan. I normally don't use outside
nzb files. I'm old fashioned and like to read headers. When I do have
an nzb file, I tend to let hellanzb handle that task.

I am online, and Pan does start to download selected tasks as I queue
them up. In this particular instance, I don't recall if I was
downloading something else with Pan at the time. I may of had the
online button unchecked. 100% positive, in the first instance I
reported, pan was online and working when I exited. Those files where
also from headers I downloaded with Pan.

No other version of Pan was installed previously, so it isn't a
problem with mismatched files in the .pan2 dir.

I went back through the group, and selected certain posts here and
there, all of which downloaded fine. There's no way I could try each
and every post. As you saw, it was quite a bit of data to download.
Normally I would never leech that much. But this was a test, so I
selected this and that just to see what would happen.

Side note -
I've been switching my main PC back and forth over the last month or
so. Being an old Slackware user, I moved to Ubuntu after Pat dropped
Gnome support. I first used Ubuntu with Warty, and stuck with Ubuntu
until just a month or so ago. Right now I'm testing out Arch 32bit and
Arch 64bit, so I won't have access to any original log files or the
like. I'm posting/reading this from a secondary Arch32bit PC without
Pan installed. Doesn't have the specs (PII 450 256ram) to handle
testing out Pan in this instance. With Ubuntu, I started to self
compile 0.121 (Ubuntu's packages at the time where out of date) with
the same compile options as the svn's I've been using with Arch. A
similar issue happened with my Ubuntu builds at least twice over the
years, but I never spent anytime looking further into it. Since at the
time pan was developed weekly, and the issue didn't occur often,
figured it was me, or already fixed.

I hope Charles is well. And personally don't mind the break that is
happening. Pan has moved so far forward, plus Transmission is, yet,
another fine piece of software.