Re (2): Dud: ISP so send mail directly?

Does that mean the ISP can determine if a clients 'stream' is going
to: http or NTTP or <email> ?

But then these detestible(sp) web-based mails would 'escape'?

I was pleased, that based on feedback from this thread, I got
`mutt` to read my gmail, which is even better than `lynx` or
`links`.

But now I realise that to SEND mutt, just uses `sendmail`.
So I'm back where I started. And I'm not convinced that
if I sweat-blood to get sendmail: smart-host working,
it will be able to fake my "From: " field to make the
mail-lists believe that I'm sending from the orininal
ISP1.

What slays me, is that whereas pop & smtp to the ISP are so
trivial, that you can just write down the steps from the RFC,
and interactively paste them as a telnet session.

`sendmail` seems to be as absurd as the-english-language.

Why doesn't somebody who claims to understand it, do a
hello-world/successive-refinment evolutionary test-sequence/
tutorial. eg:--
* send from root to non-root-user on the same PC.
* send from non-root-user to root user on the same PC.
I tried that and traced that my text had gone to /var/*,
but couldn't 'pop' it, to get a full 'cycle'.
* CONFIRM receipt ....
* send from root to ISP port 25
* send from non-root-user to ISP port xyz <-probe
* etc.

With a list of confirmatory traces for each stage.

What RFC should I fetch?

==TIA.