[Nmh-workers] another message corruption bug

hi --

sorry to have to report another corruption issue.  inc corrupts the
DKIM-Signature header in the attached message from:

DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;
        h=x-received:sender:content-type:mime-version:subject:from
        :in-reply-to:date:cc:message-id:references:to:x-mailer;
        bh=9RvGuD7z0t+zYV3baoa1XJT3sSAnFive2ajfw7XzoAI=;
        b=oegx7thD9cv2dxX30/3vHXKritCxhrTIMIsAX0cBnt1C6JXWgq7sKD5M+w/+CrkJp1
        JqUHyVQnwSqMu03WfNBYdzAVrNc7Ask11AVp+GrwniRFfW6yGXzUh/okC/fAwS4U46sh
        tAeflp3+SiSKLuo6U+00jjH8ekhtYxmWSrXBvYHQgvsyqHd5CHvPThO4qxDUJ0Zx/4mj
        sU1fI9N5mBm27WkHFgFJbBgeMZViP9LDw3wapxHK0ABucKdEM6cqOrNn+cZVuVfysNXl
        53IWIWcIOMBwL5zlG6k7pdNm3OtCnn59RML9EIPfDv+Vuk8VfHfiP7Siz7kqPQVf+8ol
        /6QQ==

to: 

DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;
        h=x-received:sender:content-type:mime-version:subject:from
        :in-reply-to:date:cc:message-id:references:to:x-mailer;
        bh=9RvGuD7z0t+zYV3baoa1XJT3sSAnFive2ajfw7XzoAI=;
        b=oegx7thD9cv2dxX30/3vHXKritCxhrTIMIsAX0cBnt1C6JXWgq7sKD5M+w/+CrkJp1
        JqUHyVQnwSqMu03WfNBYdzAVrNc7Ask11AVp+GrwniRFfW6yGXzUh/okC/fAwS4U46sh
        tAeflp3+SiSKLuo6U+00jjH8ekhtYxmWSrXBvYHQgvsyqHd5CHvPThO4qxDUJ0Zx/4mj
        sU1fI9N5mBm27WkHFgFJbBgeMZViP9LDw3wapxHK0ABucKdEM6cqOrNn+cZVuVfysNXl
53IWIWcIOMBwL5zlG6k7pdNm3OtCnn59RML9EIPfDv+Vuk8VfHfiP7Siz7kqPQVf+8ol
        /6QQ==

(i don't know whether the issue is this header, or something previous.)

i went back to the commit before f6e6ec96c9e179f7817fca4c8c22bc2bd4e417e8,
which introduced the corruption previous issue i found, but rather
than fix the problem, the inc built that commit simply dumps core. ;-)

so:

somewhere between
    commit f6e6ec96c9e179f7817fca4c8c22bc2bd4e417e8 (master~193^2~6^2~31)
    Author: David Levine <levinedl(_at_)acm(_dot_)org>
    Date:   Sun Jan 13 11:08:28 2013 -0600

      Added bytes_read to m_getfld() buffer state.  This is the
      next step in supporting ftell()/fseek().
and
    commit 55f65ae2d3baf60396d3359db952460939de03ca
    Author: David Levine <levinedl(_at_)acm(_dot_)org>
    Date:   Sun Apr 14 10:47:31 2013 -0500

        Moved #include of signal.h to h/signals.h.  And it was already
        in h/nmh.h.

the behavior changed from dumping core to losing a tab character.  i
didn't bisect further -- it's not clear that bisecting between two failures
would be all that productive.  ;-)

paul
----------------------
 paul fox, pgf(_at_)foxharp(_dot_)boston(_dot_)ma(_dot_)us (arlington, ma, 
where it's 61.7 degrees)