Hello,
thanks for all the suggestions and interest about
my question.
I thought it over a bit more, and maybe the solution
is simpler than we assumed.
What we would like to have is same mutt functionalities
on webmail interfaces, and configure these just uploading
our home muttrc file (i.e. no double email config).
This doesn't mean running mutt remotely and interactively
through some browser based terminal emulation (via Java,
JavaScript, whatever).
It would be more than enough, and far less bandwidth consuming,
something like this:
1) WWW interface is just as any other webmail interface,
html/javascript/php/frames/CGI forms/etc..
2) You compose through standard CGI form: clicking send button
invokes a standard CGI program that simply does:
cat messsage_txt| mutt -s subject \
-config_file MY_USER_NAME_MUTTRC \
-attach a_file_uploaded_with_the_form \
-recipient recipients_inserted_in_the_form
-whatever_other_options_mutt_accepts
3) at that point our standard mutt starts with the rc_file
we uploaded once, at registration time, sends the message
with the same hooks we use from home, and exits.
4) In the browser we just see some message saying "yes, message sent"
or whatever appropriate.
5) When it comes to displaying messages, or saving them to specific
folders, some php or other thing (not mutt) generates an html pages
as usual, but formatting it (as in which fields to display, how to
sort/thread/color, etc..) using what it reads from our muttrc file
on the server.
6) When we click the standard CGI "save message" button, some other
vanilla CGI thingy reads the save-hook statements in the rc file,
and saves the message as requested.
While the above suggestions represent just about the top of my programming skills
(meaning I wouldn't know how to code this),
it seems to me much simpler than we discussed yesterday, doesn't
it?
Hopefully waiting for some real hacker that implements it,
Ciao,
Marco Fioretti
