My God ... Its full of epsilons!

[Richard Lamson <rsl%max-fleischer.ila-sf.dialnet.symbolics.com@RELAY.CS.NET>]

    Date: Sat, 2 Sep 89 01:31:35 CDT
    From: "BUCKMAN%ALAN.LAAC-AI.Dialnet.Symbolics.COM"%ALAN.kahuna.DECNET.LOCKHEED.COM@Warbucks.AI.SRI.COM

	Date: Fri, 1-Sep-89 19:13:17-PDT

	Date: Fri,  1 Sep 89 19:42:07 EDT
	From: MILLER@vax.cam.nist.gov

	Mabry's message w/code reminded me of something thats been bugging me
	and I haven't looked terribly hard for a solution.  It's nothing wrong
	with Mabry or Mabry's code, its just
	 `My god...Its full of epsilons!!'

    So, about 6-12 months ago, I tried to get this solved by Symbolics
    customer service.  (i.e. the same thing happens here, and its a real
    pain in the but.)

Sounds to me like some host along the way is doing something to your
message to cause it to lose some essential piece of information.  Zmail
knows how to interpret those pesky epsilons correctly when they (and the
contents of the header) are transmitted correctly.  Of course, the
epsilon character is octal 6, outside of the normal "printing ASCII"
character set for which SMTP is well-defined, which means that an SMTP
server is permitted to drop them on the floor.  However, from your
description it sounds like the epsilons themselves are correctly sent.

One other possibility is that your mailer stores the mail in some format
which does not exactly preserve the contents of the mail.  For example,
if you read your mail file from a VMS host, Zmail may be attempting to
interpret VMS' format of mail file, and the format may not be capable of
storing the headers properly.

Here is a copy of the header information which was in the mail when I
received it from Mabry:

Character-Type-Mappings: (1 0 (NIL 0) (NIL :BOLD NIL) "CPTFONTCB")
                         (2 0 (NIL 0) (NIL :ROMAN NIL) "CPTFONT")
                         (3 0 (NIL 0) (NIL :ITALIC NIL) "CPTFONTI")
                         (4 0 (NIL 0) (:SWISS :BOLD :SMALLER) "HL10B")
                         (5 0 (NIL 0) (:SWISS :BOLD :SMALL) "HL10B")
Fonts: CPTFONT, CPTFONTCB, CPTFONT, CPTFONTI, HL10B, HL10B

You should verify:

1.  That the above is what was in your mail (not, e.g., "X-FONTS" or
    something where some SMTP server which knew better than you do
    decided to rewrite the headers).

2.  That your Zmail is parsing those headers correctly (the message
    display should not say "Bad Header" in the mode window of your Zmail
    frame; if it does, say m-X Explain Bad Header).

If you are receiving your mail on some civilized host (e.g., a lisp
machine mailer) and reading it with Zmail, and neither of the above
conditions is false, I propose we try the following experiment offline:
I will send you a message with character styles in it.  Assuming Zmail
doesn't display the character style changes correctly, you will use the
text editor on your incoming mail file, pick out the contents of my
message (that file SHOULD contain epsilons), replace all the epsilon
characters with some ASCII character which is not in the message (e.g.,
#\%), and mail it back to me.  We'll see if we can find out what's
happening to the message between here and there.

[Aside: An experiment you can try immediately: send yourself some mail
with character-style changes in it; if you have two different mail paths
to yourself (e.g., directly to your lispm mailer and via a forwarding
address on your local Decnet gateway), send it to yourself via both
paths.  Look at your incoming mail file to see if you can pick out any
differences.

A useful tool for this is Zmacs command m-X Compare Windows.  Enter
two-window mode (c-X 2), put the two different versions of the text in
the two windows (use c-X O (letter o) to change windows), put the cursor
at the logical beginning of each version, and say m-X Compare Windows.
Zmacs will move both cursors to the place where the two windows first
differ.]