Re: [Pan-users] Re: 'hundreds' of strange extracted file when downloadingmultipart messages

[wim . delvaux]

Thanx for the explanation.

Where did I 'accidently' ask to save the 'text' messages too ?

Thx

W

On Thursday 24 April 2008 18:01:25 Duncan wrote:
> Travis <address@hidden> posted
> address@hidden, excerpted below, on  Thu, 24
>
> Apr 2008 06:57:44 -0700:
> > ----- Original Message -----
> >
> >> I run pan version 0.132.
> >>
> >> When I download a multipart message from news.usenetserver.com I get a
> >> lot of files with names like :
> >>
> >> M8WPj.14920$HY2.13951-epkf52SaZeZP2roSf1MJ43itZ/
> >> address@hidden
> >>
> >> I got other extension too (so not only with usenetserver.com.msg).
> >
> > You have Pan set to save "text" as well as the attachments.
>
> Travis is correct, depending on where you are looking, but here's a
> somewhat more detailed explanation.
>
> Pan saves messages using the (sanitized to filesystem safe) Message-ID,
> which according to the RFCs should be globally unique -- there should
> never be two different messages anywhere in the world at any time with
> the same Message-ID.
>
> Different news clients (and original posting servers, if the news client
> doesn't supply its own) use different algorithms to come up with the
> Message-ID, however.  Many use some form of a combination of from address
> (either the poster's, or the posting server's) to hopefully keep
> collisions in "space" from occurring, timestamp, to keep collisions in
> time from occurring, and random number, just in case, but it's up to the
> implementation exactly what they use, as long as it's unique.
>
> That's why the message files appear to have such strange names -- they
> are based on the Message-ID, which isn't standardized except in the
> characters it may contain and that it must be unique, and isn't really
> designed for file-naming, the use to which pan puts it.  Pan uses the
> Message-ID as a filename, however, in ordered to help keep things
> straight, since it's unique to the message and pan is designed to be able
> to track and know when it has already downloaded the same message across
> multiple servers.  (The other form of message numbering, normally
> sequential by group, is server-specific, and therefore won't help
> tracking messages between servers.)
>
> All that said, normally, the only place these filenames appear is in
> pan's cache, which by default is pretty small, 10 MB. (The cache can be
> set larger manually, by editing preferences.xml in pan's settings dir
> directly. I run a multi-gigabyte cache, save to cache, and then sort and
> save off messages from there after they are all stored in local cache.)
> Since pan handles its cache automatically, you'll not normally see these
> files unless you go manually trawling thru the cache.
>
> However, if as Travis suggested, if you tell pan to save the text message
> itself, not just attachments, it'll save this more or less raw text
> message.  This is nice if you want to save a text message, tho you may
> want to rename it to something more appropriate (say the subject)
> afterward.  I do this from time to time.  It's also useful on "broken"
> attachments that pan can't properly save on its own, however, as it may
> be possible to recover the attachment using other tools designed to deal
> for the purpose.  I do this too, occasionally.
>
> So basically, if you are telling pan to save the text message not just
> the attachments, you'll get these wherever you have pan saving its
> messages.  Tell it to save attachments only, and you'll not have to worry
> about them.  Or, if you are examining pan's cache, don't worry about it,
> as pan should manage them on its own unless you've deliberately set it up
> to handle manually, as I have.