sieve-manage suggestions

sieve-manage suggestions

[Kai Großjohann]

I think that sieve-manage should be autoloaded.

Also, almost everything can be submitted with C-c C-c, but not in a
sieve buffer.  How about having C-c C-c do like C-c C-l and then kill
(or bury?) the buffer?  (I think that comment-region is not all that
important; now that we have comment-dwim.)

How about getting connection info from the Gnus servers?  I have to
enter the Sieve server and my user name and my password, but Gnus (and
nnimap) knows all this already.
-- 
Ambibibentists unite!

Re: sieve-manage suggestions

[Kai Großjohann]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> Also, almost everything can be submitted with C-c C-c, but not in a
> sieve buffer.  How about having C-c C-c do like C-c C-l and then kill
> (or bury?) the buffer?  (I think that comment-region is not all that
> important; now that we have comment-dwim.)

Of course, making C-x C-s do the right thing would be super-nifty.
But it's work.

I now have one Sieve script shown my sieve-manage, and it's called
template.siv.  Hm.  It's kinda strange to have this as something that
does real work :-)  I don't quite remember how I got this, but I
think I did only obvious things.  So how about arranging it so that
the Sieve script gets a better name when dumb users like me do the
obvious things?
-- 
Ambibibentists unite!

Re: sieve-manage suggestions

[Simon Josefsson]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>
>> Also, almost everything can be submitted with C-c C-c, but not in a
>> sieve buffer.  How about having C-c C-c do like C-c C-l and then kill
>> (or bury?) the buffer?  (I think that comment-region is not all that
>> important; now that we have comment-dwim.)
>
> Of course, making C-x C-s do the right thing would be super-nifty.
> But it's work.

I thought about using C-x C-s but I was afraid that users might expect
too much if that is used -- e.g., autosave, crash backups,
save-places.

> I now have one Sieve script shown my sieve-manage, and it's called
> template.siv.  Hm.  It's kinda strange to have this as something that
> does real work :-)  I don't quite remember how I got this, but I
> think I did only obvious things.  So how about arranging it so that
> the Sieve script gets a better name when dumb users like me do the
> obvious things?

What would a good solution be?  Simply asking the user for a proper
name on upload iff the name equals "template.siv"?

Re: sieve-manage suggestions

[Kai Großjohann]

Simon Josefsson <jas@extundo.com> writes:

> I thought about using C-x C-s but I was afraid that users might expect
> too much if that is used -- e.g., autosave, crash backups,
> save-places.

Indeed.

>> I now have one Sieve script shown my sieve-manage, and it's called
>> template.siv.  Hm.  It's kinda strange to have this as something that
>> does real work :-)  I don't quite remember how I got this, but I
>> think I did only obvious things.  So how about arranging it so that
>> the Sieve script gets a better name when dumb users like me do the
>> obvious things?
>
> What would a good solution be?  Simply asking the user for a proper
> name on upload iff the name equals "template.siv"?

That's possible.  Or maybe choose another name?  Or use auto-insert
to automatically insert some stuff when the user clicks on `new'?

I'm having some trouble understanding this business of multiple
files, only one of which can be active.  Why?  Hm.
-- 
Ambibibentists unite!

Re: sieve-manage suggestions

[Simon Josefsson]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> I think that sieve-manage should be autoloaded.

It should relady be, are you loading gnus-load?

> Also, almost everything can be submitted with C-c C-c, but not in a
> sieve buffer.  How about having C-c C-c do like C-c C-l and then kill
> (or bury?) the buffer?  (I think that comment-region is not all that
> important; now that we have comment-dwim.)

Good idea, added.

> How about getting connection info from the Gnus servers?  I have to
> enter the Sieve server and my user name and my password, but Gnus (and
> nnimap) knows all this already.

The current situation is not very user friendly, no.  But some
problems:

1) The sieve package is not Gnus specific.  This means the proper
   solution is to write a generic server-username-password interface
   which other packages may use too, but it sounds like boring work
   that really ought to be solved by a generic package that can be
   used by other elisp packages as well.

2) There really isn't anything saying that the sieve servername or
   username or password is the same as the IMAP dittos.  OTOH, they
   make for a very good default, which means the generic
   server-username-password handling package above should deal with
   suggested defaults as well.

A simple hack that makes sieve default to the current Gnus server and
uses .authinfo for username/passwords would go a long way of improving
the current situation though.  Added to my todo list...

Re: sieve-manage suggestions

[Kai Großjohann]

Simon Josefsson <jas@extundo.com> writes:

> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>
>> I think that sieve-manage should be autoloaded.
>
> It should relady be, are you loading gnus-load?

I have no idea.  Should I?  I mean, when I start Emacs, Gnus is
already autoloaded: I type M-x gnus RET and there I am.

> A simple hack that makes sieve default to the current Gnus server and
> uses .authinfo for username/passwords would go a long way of improving
> the current situation though.  Added to my todo list...

Hm.  Hmmm...  Actually, I think it would be sufficient to have a
variable sieve-manage-default-server...

And then it would be cool if sieve could look up things in
~/.authinfo.  But I have no idea how to make it so that nnimap and
sieve find different passwords for the same server.  Hm.  Maybe the
port number is sufficient for that?
-- 
Ambibibentists unite!

Re: sieve-manage suggestions

[Simon Josefsson]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>>
>>> I think that sieve-manage should be autoloaded.
>>
>> It should relady be, are you loading gnus-load?
>
> I have no idea.  Should I?  I mean, when I start Emacs, Gnus is
> already autoloaded: I type M-x gnus RET and there I am.

The Gnus that is autoloaded is probably the Gnus shipped with Emacs,
to load the autoload cookies for Oort you need some extra things:

** (require 'gnus-load)

If you use a stand-alone Gnus distribution, you'd better add (require
'gnus-load) into your ~/.emacs after adding the Gnus lisp directory
into load-path.

File gnus-load.el contains autoload commands, functions and variables,
some of which may not be included in distributions of Emacsen.

>> A simple hack that makes sieve default to the current Gnus server and
>> uses .authinfo for username/passwords would go a long way of improving
>> the current situation though.  Added to my todo list...
>
> Hm.  Hmmm...  Actually, I think it would be sufficient to have a
> variable sieve-manage-default-server...
>
> And then it would be cool if sieve could look up things in
> ~/.authinfo.  But I have no idea how to make it so that nnimap and
> sieve find different passwords for the same server.  Hm.  Maybe the
> port number is sufficient for that?

Yup, this doesn't sound too hard.

>>> I now have one Sieve script shown my sieve-manage, and it's called
>>> template.siv.  Hm.  It's kinda strange to have this as something that
>>> does real work :-)  I don't quite remember how I got this, but I
>>> think I did only obvious things.  So how about arranging it so that
>>> the Sieve script gets a better name when dumb users like me do the
>>> obvious things?
>>
>> What would a good solution be?  Simply asking the user for a proper
>> name on upload iff the name equals "template.siv"?
>
> That's possible.  Or maybe choose another name?  Or use auto-insert
> to automatically insert some stuff when the user clicks on `new'?

Hm.  Yes.

> I'm having some trouble understanding this business of multiple
> files, only one of which can be active.  Why?  Hm.

Because it is how the Sieve Manage protocol works. :-)

I guess the intention is to allow people to have several scripts ready
for use on the server; one for holidays, one for business travel, one
for normal usage etc...  If these where stored on the client, you'd
need to bring a copy of the script everywhere you go (if you want to
change them).

Re: sieve-manage suggestions

[Kai Großjohann]

Simon Josefsson <jas@extundo.com> writes:

> The Gnus that is autoloaded is probably the Gnus shipped with Emacs,
> to load the autoload cookies for Oort you need some extra things:
>
> ** (require 'gnus-load)
>
> If you use a stand-alone Gnus distribution, you'd better add (require
> 'gnus-load) into your ~/.emacs after adding the Gnus lisp directory
> into load-path.

Wah :-( So long I've been using the CVS version and I still didn't
notice this.  I've added this now.

>> I'm having some trouble understanding this business of multiple
>> files, only one of which can be active.  Why?  Hm.
>
> Because it is how the Sieve Manage protocol works. :-)
>
> I guess the intention is to allow people to have several scripts ready
> for use on the server; one for holidays, one for business travel, one
> for normal usage etc...  If these where stored on the client, you'd
> need to bring a copy of the script everywhere you go (if you want to
> change them).

Yes, but /why/?  Most of my script is still needed even when I'm on
holidays or travel.  And the single "vacation" statement can easily
be commented out or something.

But then, these things are supposed to be for nonprogrammers who
might have a different opinion about it.  But for me personally, I
don't see the point.
-- 
Ambibibentists unite!

Re: sieve-manage suggestions

[Simon Josefsson]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

>>> I'm having some trouble understanding this business of multiple
>>> files, only one of which can be active.  Why?  Hm.
>>
>> Because it is how the Sieve Manage protocol works. :-)
>>
>> I guess the intention is to allow people to have several scripts ready
>> for use on the server; one for holidays, one for business travel, one
>> for normal usage etc...  If these where stored on the client, you'd
>> need to bring a copy of the script everywhere you go (if you want to
>> change them).
>
> Yes, but /why/?  Most of my script is still needed even when I'm on
> holidays or travel.  And the single "vacation" statement can easily
> be commented out or something.

Perhaps it is meant for people that doesn't edit Sieve scripts
directly as text.  E.g. they simply have three scripts "holiday" and
"travel" and "normal" and switch between which one is active.
Starting to edit each of the scripts and disabling or enabling certain
parts of the file depending on which mode you are in might be rather
difficult in a limited GUI rule system.

But I'm only guessing, I only ever use one script too.  (Well,
actually, I do have a few others, but they are named "test", "test2"
etc, and were created only in order to test the protocol commands...)

Do you think there is something here that can be translated into an
improvement for emacs-sieve users?

Re: sieve-manage suggestions

[Kai Großjohann]

Simon Josefsson <jas@extundo.com> writes:

> Do you think there is something here that can be translated into an
> improvement for emacs-sieve users?

Maybe it means that people only ever use one script anyway?  So the
script could have a nice name, such as default, and the NEW button
could be hidden a bit.  Right now it's really staring you in the
face.  (Maybe just put it at the bottom?)
-- 
Ambibibentists unite!