Re: [Pan-users] Multipart pic display

[Per Hedeland]

Darren <address@hidden> wrote:
>
>Per Hedeland wrote:
> > Hm, the overwhelming response to that made me go looking for a fix
>> myself:-) - patch for 0.106 below. The problem was that the handling of
>
>I actually meant to take a look and see if the occasional issue I see of
>pan seeing some attachments as incomplete that other newsreaders see
>properly is this same problem, so your post wasn't completely ignored.  ;-)

Thanks!:-)

>I will try this patch laster tonight and see if it works for me, thanks!

OK, though I suspect it won't, if I understand correctly what you mean
with "incomplete" attachments: In the case where my fix applies, pan
thinks it has found all the parts (and it has), i.e. it shows the
"complete puzzle" icon rather than the "single piece" one - it is *only*
a problem with the rendering in the body pane (which of course is only
done when pan "thinks" it makes sense, and incidentally uses mechanisms
that are entirely separate from the ones used for "save attachments").

IME, the typical case where pan thinks it hasn't got all the parts
("single piece" icon), even though they're actually there, is when it
gets confused about the "(1/4)", "[1 of 5]", etc hints in the Subject
line that decoding of multi-part postings in News unfortunately has to
rely on (MIME has well-defined methods to describe such messages, but it
seems it just won't catch on for News).

E.g., there may be a suite of posts like:

Subject: Mumble foo.jpg (1/3) [1 of 4]
Subject: Mumble foo.jpg (2/3) [1 of 4]
Subject: Mumble foo.jpg (3/3) [1 of 4]
Subject: Mumble bar.jpg (1/2) [2 of 4]
Subject: Mumble bar.jpg (2/2) [2 of 4]
...

- and in at least some such cases I've seen (old) pan mistakenly think
that the "[n of N]" rather than the "(n/N)" item is the "part indicator"
- IIRC, (old) pan scanned the Subject line from the end backwards and
picked the first "indicator" it found. There is no obviously correct way
to deal with this mess, but based on a very un-scientific survey, it
seemed to me that when both "(*)" and "[*]" indicators were present,
"(*)" was more likely to be the "right" one, and I did a patch
accordingly (i.e. pick "[*]" only if there is no "(*)"), which seemed to
improve things.

Hm, looking at that code in new pan, I see that it has gotten a couple
of improvments - it now requires the separator to be "/" or "|", i.e. an
item using "of" is ignored, and there is a sanity check that n <= N - so
this may not be the problem you're seeing. But if it seems it could
match (e.g. the Subject line has both "(n/N)" and "[n/N]", with "(n/N)"
being the right indicator), I could try and forward-port my old patch
for "(*)"-preference.

--Per