Re: a short introduction to Gnus vs. Spam with spam.el

[Ted Zlatanov <tzz@lifelogs.com>]

On Fri, 17 Jan 2003, Derrell.Lipman@UnwiredUniverse.com wrote:
> Ted Zlatanov <tzz@lifelogs.com> writes:
> 
>> At http://lifelogs.com/spam/ 

> Just a few comments:
> 
> - The diagram in the introduction shows Spam and Ham going into the
>   MTA, but the Concepts section defines Spam and Ham as distinct
>   from "undetermined messages" Shouldn't the diagram really show
>   just "Undetermined Messages" (possibly with Spam/Ham in
>   parenthesis or something)?

Messages are never "undetermined" except as far as Gnus is concerned,
that's what I was trying to show.  Mail is always either spam or ham
(but one man's spam is another's ham :)

> - In Spliting Incoming Mail, spam-use-ifile says that the user has a
>   choice of using ifile for all classification or using it only for
>   spam, but there's no indication (at least not in this section) of
>   how to specify that choice.

I would prefer to leave such details for the Gnus manual.  The
presentation is meant to be an introduction, so I'd like to keep it
brief and focused on explaining to new users why they want Gnus +
spam.el.

> - While reading this, I'm asking myself why I can't use multiple
>   spam splitters and processors.  Is there some reason not to use
>   bogofilter and blacklist, for example?  It seems from the early
>   text, that the user must choose one.  The later screen shots
>   indicate that multiple processors may be checked.

The answer is yes and no.  It wouldn't make sense to use ifile and
bogofilter together, for instance.  Also, ifile won't coexist well
with the other splitters if it is used to split into all folders, not
just spam/non-spam.  I didn't want to bring up those issues in the
presentation.

> - 'H' as a spam mark seems counter-intuitive when the opposite is
>   Ham.  (I realize that 'S' isn't available, but this is worth some
>   consideration.)

I know, I know.  Fortunately it's just a character, and can be
customized.  I tried hard to eliminate direct dependence on the 'H'
character in spam.el, but it could be a problem depending on how Gnus
stores old marks, and whether there's any other code that doesn't use
the spam-mark as a symbol but directly as a character value.  I would
assume that won't be a problem.

So, the question is: what should the new "spam" mark be?  "$" seems an
obvious choice; I personally don't like the way it looks in the
summary buffer with a monospaced font but that's hardly crucial.
Also, "$" is not as meaningful in countries that don't use dollars.  I
would love to use an extended Unicode character, but that will annoy
everyone.  So maybe "@" would be good?  What do you (the spam.el
current and potential users) think?

Ted