Re: [courier-users] E-Mails rejected due to MIME boundary too long

[Sam V. <mr...@co...>]

Helge Kreutzmann writes:

> Hello Sam,
> first of all, sorry for the long delay. My admin was absent so I could
> not try it out earlier.
>
> On Tue, May 01, 2012 at 06:22:45PM -0400, Sam Varshavchik wrote:
> > Helge Kreutzmann writes:
> >
> > >I tried searching for a knob to shorten the MIME boundary, but was
> > >unsucessfully (also in the list archive).
> >
> > The first part of the boundary is your system's hostname. Set your
> > system hostname to a shorter name.
>
> I don't think this is possible on this machine, unfortuantely.
>
> > Try also to change the formatting string in rfc2045mkboundary.c to:
> >
> >     sprintf(p, "=_%-30s-%s-%s-%s", hostnamebuf,
> >        pidbuf, timebuf, cntbuf);
> >
> > Not ideal, but this is the easiest fix.
>
> I rebuild with this fix and at first it seemed to work, some test
> mails went through:
> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp- 
> signature"; boundary="=_static.213-239-213-133.clients.your- 
> server.de-2710-1337109850-0001-2"

Try:

     sprintf(p, "=_%-1.30s-%s-%s-%s", hostnamebuf,
        pidbuf, timebuf, cntbuf);

Also, a similar change to gpglib/gpg.c in newboundary():

	sprintf(buffer, "=_%-1.30s-%u-%u-%04u",
		hostnamebuf, (unsigned)getpid(),
		(unsigned)t, ++counter);
	return (buffer);