Re: Improper handling of some mime types

[Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann)]

Kurt Swanson <direct@reply.not.desired> writes:

> So you're saying I essentially have to hit `K b' in every single
> article I read just in case there's (by Gnus) hidden data, and then,
> if there is an html version, choose that, and the click all buttons,
> hoping one of them will do something useful for me?

> To me this is a *serious* error.  There are two mime parts in
> message-id: tQCa8.5559$qt6.469306@newsread2.prod.itd.earthlink.net in
> the alt.binaries.pictures.orchid newsgroup.  To put it simply, there
> is part 1a, 1b, and 2; i.e. there are two versions of part one (they
> share the same id code) and one part 2 (different id).  Part two
> should be recognized as separate (and equal?) and displayed,
> regardless of the user's choice of a or b for part 1.  The fact that
> part 2's very existence can only be seen (and then handled poorly)
> when part 1b is chosen can't be the desirable functionality here...
>
> Please go and check out this article yourself...

I can't do that, for I can't find a server hosting that group.
Sorry.  (I tried Google, did I do something wrong there?)

Whether Gnus is in error depends on the structure of the message.  It
could be like this (using pseudo-XML):

    <part type="multipart/alternative">
      <part type="text/plain"/>
      <part type="multipart/related">
        <part type="text/html"/>
        <part type="image/png"/>
      </part>
    </part>

It could also look like this:

    <part type="multipart/mixed">
      <part type="multipart/alternative">
        <part type="text/html"/>
        <part type="text/plain"/>
      </part>
      <part type="image/png"/>
    </part>

Your description sounds like the first variant, and then the Gnus
rendering would be correct.

kai
-- 
~/.signature is: umop 3p!sdn    (Frank Nobis)