From: "Jeremy Blackman" <l...@maison-otaku.net>
Subject: Re: Problem attaching files of type Text/HTML
Date: 1999/07/25
Message-ID: <932900717.626.40@news.remarQ.com>#1/1
X-Deja-AN: 505050614
References: <Pine.OSF.4.10.9907200951370.12764-100000@goedel3.math.washington.edu>
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3155.0
X-Complaints-To: newsabuse@remarQ.com
X-Trace: 932900717.626.40 3Y9WTANSL04F0D87AC qube-01.us-ca.remarq.com
Organization: Posted via RemarQ, http://www.remarQ.com - The Internet's Discussion Network
NNTP-Posting-Date: Sun, 25 Jul 1999 11:05:17 GMT
Newsgroups: comp.mail.pine

>  I reported the same problem a few weeks ago, I received a message from
>the pine team asking me for my .pine-debug files, and they answered me that
>this one was a difficult one, they hadn't catch it yet. Up to the moment no
>one has answered me why this occurs, if someone lets you know, please let
>me know. I myself tried several ways to change the headers, and could not
>do it.


Hey, wait, me and one other person not only tracked this down but SUBMITTED
A FIX several weeks ago.  <grumble>

It's not, technically speaking, a bug in PINE.  It is a bug in egcs/gcc,
where in one particular case, an optimization causes a return result from
one of the MIME handler routines to get utterly clobbered.  (This is
probably also why it does BASE64 encoding when quoted-printable would be
more appropriate, but we didn't bother to try debugging that one.
Step-debugging PINE for one bug is pain enough, thank you.)

I know for a fact that this bug affects x86 versions of egcs/gcc after 2.7.2
(because I tried it on a variety of machines), and I've been told by someone
that the Alpha architecture also has this problem.  (The bug was submitted
to the egcs/gcc team, as well.)  The bug causes the return to contain a '0',
no matter what the actual return result was, when the program returns to the
calling function - this causes PINE to assume certain (incorrect) defaults.

With a simple alteration to that particular routine in PINE to force
egcs/gcc to be unable to optimize that one step (e.g. add 'sleep(0);' before
the return), PINE correctly attaches things again.  This particular bug
affects several MIME types, text/html is merely the most noticable.

The patch in question can be found at:
ftp://ftp.listar.org/pub/listar/other/pine4.10-mimetype.patch

(The bug was discovered while working on the MIME handling for the Listar
mailing list package, hence the location of the patch.  There's also a patch
to add minimal RFC2369 support to PINE in that same directory, as
pine4.10-listserv.patch)

Hope that helps!