Problems with 5.1.1 milter

[Mark Sapiro]

Shawn Iverson wrote:

> There's a performance fix in this commit that should help with most of
> this.  The crashing is likely from partially written messages in the
> milterin queue as a result of not writing the file in a single pass.  I
> hope to get a few more issues resolved before doing another release.
> 
> https://github.com/MailScanner/v5/commit/5083b4e0f0eecbec84a5ab94538b0ffbe35f90f1


I've been doing a bit more testing. I installed the above change, and
even with that I see this issue.

I submit a message whose headers end with

MIME-Version: 1.0
Content-Type: multipart/alternative;
        boundary="b1_f5bca29580e2b9cb455f487aa1a5bd28"

And the resultant message after MailScanner has headers ending with

MIME-Version: 1.0
Content-Type: multipart/alternative;
X-msapiro-MailScanner-ID: 12CA95A5DC.AD7FE
X-msapiro-MailScanner: Found to be clean
X-msapiro-MailScanner-SpamCheck: not spam, SpamAssassin (not cached,
        score=1.132, required 6, DKIM_SIGNED 0.10,
        HEADER_FROM_DIFFERENT_DOMAINS 0.00, MIME_NO_TEXT 1.00,
        RCVD_IN_DNSWL_NONE -0.00, T_DKIM_INVALID 0.01, T_MIME_MALF 0.01,
        T_TVD_MIME_NO_HEADERS 0.01, UNPARSEABLE_RELAY 0.00)
X-msapiro-MailScanner-SpamScore: s
X-msapiro-MailScanner-From: mark at msapiro.net
X-Spam-Status: No
        boundary="b1_f5bca29580e2b9cb455f487aa1a5bd28"

I.e. the MailScanner report headers are inserted before the last line of
the headers rather than appended after. This is not always a problem,
but in this case, it causes the 'boundary=' to be separated from the
'Content-Type:' which is a problem.

-- 
Mark Sapiro <mark at msapiro.net>        The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan