Do not want agantized, nnml backend.

Do not want agantized, nnml backend.

[小関 吉則]

[-- Attachment #1: Type: text/plain, Size: 234 bytes --]

I'm using gnus-agent and nnml backend.

(setq gnus-select-method '(nnml ""))
(setq gnus-secondary-select-methods nil)

But when Unplugged, In Summary buffer marked %. I can't read
mails.

Agentized backend is Only nntp and nnimap???


[-- Attachment #2: Type: text/x-patch, Size: 1570 bytes --]

RCS file: /usr/local/cvsroot/gnus/lisp/ChangeLog,v
retrieving revision 6.1487
diff -u -r6.1487 ChangeLog
--- ChangeLog	2002/08/22 11:54:06	6.1487
+++ ChangeLog	2002/08/26 02:36:44
@@ -1,3 +1,9 @@
+2002-08-26  KOSEKI Yoshinori  <kose@meadowy.org>
+
+	* gnus-agent.el (gnus-agent-agentized-backend): New variable.
+	(gnus-agentize): Agentized, if When gnus-select-method is
+	gnus-agent-agentized-backend.
+
 2002-08-22  Kai Großjohann  <Kai.Grossjohann@CS.Uni-Dortmund.DE>
 
 	* gnus-art.el (gnus-emphasis-alist): Strikethru had a lot of false
Index: gnus-agent.el
===================================================================
RCS file: /usr/local/cvsroot/gnus/lisp/gnus-agent.el,v
retrieving revision 6.73
diff -u -r6.73 gnus-agent.el
--- gnus-agent.el	2002/06/16 21:13:49	6.73
+++ gnus-agent.el	2002/08/26 02:36:45
@@ -156,6 +156,8 @@
 (defvar gnus-agent-send-mail-function nil)
 (defvar gnus-agent-file-coding-system 'raw-text)
 (defvar gnus-agent-file-loading-cache nil)
+(defvar gnus-agent-agentized-backend '(nntp nnimap)
+  "Agentized, When these backends.")
 
 ;; Dynamic variables
 (defvar gnus-headers)
@@ -418,7 +420,8 @@
 					 message-send-mail-real-function
 					 message-send-mail-function)
 	  message-send-mail-real-function 'gnus-agent-send-mail))
-  (unless gnus-agent-covered-methods
+  (when (and (not gnus-agent-covered-methods)
+	     (memq (car gnus-select-method) gnus-agent-agentized-backend))
     (setq gnus-agent-covered-methods (list gnus-select-method))))
 
 (defun gnus-agent-queue-setup ()

[-- Attachment #3: Type: text/plain, Size: 17 bytes --]

-- 
こせき

Re: Do not want agantized, nnml backend.

[Simon Josefsson]

小関 吉則 (KOSEKI Yoshinori) <kose@meadowy.org> writes:

> But when Unplugged, In Summary buffer marked %. I can't read
> mails.
>
> Agentized backend is Only nntp and nnimap???

Amazing, I was about to add a patch that agentizes nntp and nnimap by
default, and it fixes your problem too.

I've committed my patch now, let me know if it doesn't work.

For everyone's information; GNUS-NEWS now says the following related
to the agent.

** gnus-agent

The Gnus Agent is now enabled by default.  This means that, e.g.,
headers are not downloaded from agentized servers by default (agentize
servers by using `J a' in the server buffer).  Gnus will not start to
download articles unless you instruct it to do so, though, by using
e.g. J u or J s from the group buffer.  Revert to the old behaviour
with `(setq gnus-agent nil)'.  Note that putting (gnus-agentize) in
~/.gnus is not needed any more.  Now, by default all nntp and nnimap
servers from gnus-select-method and gnus-secondary-select-method are
agentized by default (earlier only the server in gnus-select-method
was agentized).  You can view and change which servers should be
agentized in the Server Buffer.

Re: Do not want agantized, nnml backend.

[Dan Christensen]

Simon Josefsson <jas@extundo.com> writes:

> ** gnus-agent
>
> The Gnus Agent is now enabled by default.  This means that, e.g.,
> headers are not downloaded from agentized servers by default (agentize
              ^^^ now?
> servers by using `J a' in the server buffer).

Dan

Re: Do not want agantized, nnml backend.

[Harry Putnam]

Dan Christensen <jdc+news@uwo.ca> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> ** gnus-agent
>>
>> The Gnus Agent is now enabled by default.  This means that, e.g.,
>> headers are not downloaded from agentized servers by default (agentize
>               ^^^ now?
>> servers by using `J a' in the server buffer).

I'm a little confused here..  Which is correct?  The first (Simons')
messages went on and said a special var would have to be set before  
downloading would happen, even if a server was `agentized'.

So is the regular nntp server now *NOT* agentized by default.  And
once agentized, one must set the special var as well?...
 confused..

Didn't  agentized servers always download headers by default, so
saying `now' seems wrong. 

Re: Do not want agantized, nnml backend.

[Simon Josefsson]

Harry Putnam <reader@newsguy.com> writes:

> Dan Christensen <jdc+news@uwo.ca> writes:
>
>> Simon Josefsson <jas@extundo.com> writes:
>>
>>> ** gnus-agent
>>>
>>> The Gnus Agent is now enabled by default.  This means that, e.g.,
>>> headers are not downloaded from agentized servers by default (agentize
>>               ^^^ now?
>>> servers by using `J a' in the server buffer).
>
> I'm a little confused here..  Which is correct?  The first (Simons')
> messages went on and said a special var would have to be set before  
> downloading would happen, even if a server was `agentized'.
>
> So is the regular nntp server now *NOT* agentized by default.  And
> once agentized, one must set the special var as well?...
>  confused..

Is the new text less confusing?  (See last post.)

The intention is that you shouldn't do anything to have your nntp and
nnimap groups agentized.

Right now though, servers in g-s-m or g-s-s-m are not agentized.
(E.g., manually added ones in the server buffer.)  Maybe they should?

> Didn't  agentized servers always download headers by default, so
> saying `now' seems wrong. 

When you were plugged, Gnus used headers from the remote server, not
the local agent cache, IIRC.  Now, Gnus uses headers from the local
cache all the time.

Re: Do not want agantized, nnml backend.

[Simon Josefsson]

Dan Christensen <jdc+news@uwo.ca> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> ** gnus-agent
>>
>> The Gnus Agent is now enabled by default.  This means that, e.g.,
>> headers are not downloaded from agentized servers by default (agentize
>               ^^^ now?
>> servers by using `J a' in the server buffer).

I think "not" was correct (headears are not downloaded from servers,
they are retrieved from the agent cache), but it was rather muddy.
Another try:

  The Gnus Agent is now enabled by default, and all nntp and nnimap
  servers from gnus-select-method and gnus-secondary-select-method are
  agentized by default (earlier only the server in gnus-select-method
  was agentized by the default, when the agent was enabled).  This
  means that, e.g., headers are retrieved from the Agent cache instead
  of the backends when possible.  You can enroll or remove servers
  with `J a' and `J r' in the server buffer.  Gnus will not download
  articles unless you instruct it to do so, though, by using `J u' or
  `J s' from the Group buffer.  Revert to the old behaviour with
  `(setq gnus-agent nil)'.  Note that putting (gnus-agentize) in
  ~/.gnus is not needed any more.

Re: Do not want agantized, nnml backend.

[Harry Putnam]

Simon Josefsson <jas@extundo.com> writes:

Read you answer to me [Putnam] so included your clarified comments:
> I think "not" was correct (headears are not downloaded from servers,
> they are retrieved from the agent cache), but it was rather muddy.
> Another try:
>
>   The Gnus Agent is now enabled by default, and all nntp and nnimap
>   servers from gnus-select-method and gnus-secondary-select-method are
>   agentized by default (earlier only the server in gnus-select-method
>   was agentized by the default, when the agent was enabled).  This
>   means that, e.g., headers are retrieved from the Agent cache instead
>   of the backends when possible.  You can enroll or remove servers
>   with `J a' and `J r' in the server buffer.  Gnus will not download
>   articles unless you instruct it to do so, though, by using `J u' or
>   `J s' from the Group buffer.  Revert to the old behaviour with
>   `(setq gnus-agent nil)'.  Note that putting (gnus-agentize) in
>   ~/.gnus is not needed any more.

So, it is the case that nothing gets downloaded by default, even in
agentized groups?.  That seems wrong.

By `Revert to the old behavior' do we mean revert to downloading
agentized groups by default? Or, not having agent enabled by default?

Depending on above answers:
I don't see the advantage of not using gnus-agentize, but using 
(setq gnus-agent nil) to do what `gnus-agentize' used to do.
Especially since the syntax, on first glance, looks as if it would
shut the agent off.

Further confusion: If (setq gnus-agent nil) is set, reverting to old
behavior, does that including needing gnus-agentize at the bottom?

Seems like asking for confusion.... but maybe its just me.
Hope I'm not just nitpicking here, but it still seems pretty
confusing, but maybe because I haven't been following developments
very much for a while.

If a group is set agentized, it seems the default one would expect is
that downloading of headers would be controlled by the default
predicate.  Further categories would take it from there by user
specifying them as always.

In that scenario, many of us have used a default predicate of `false', 
shifting responsibility for downloading to custom `categories'.

Re: Do not want agantized, nnml backend.

[Simon Josefsson]

Harry Putnam <reader@newsguy.com> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
> Read you answer to me [Putnam] so included your clarified comments:
>> I think "not" was correct (headears are not downloaded from servers,
>> they are retrieved from the agent cache), but it was rather muddy.
>> Another try:
>>
>>   The Gnus Agent is now enabled by default, and all nntp and nnimap
>>   servers from gnus-select-method and gnus-secondary-select-method are
>>   agentized by default (earlier only the server in gnus-select-method
>>   was agentized by the default, when the agent was enabled).  This
>>   means that, e.g., headers are retrieved from the Agent cache instead
>>   of the backends when possible.  You can enroll or remove servers
>>   with `J a' and `J r' in the server buffer.  Gnus will not download
>>   articles unless you instruct it to do so, though, by using `J u' or
>>   `J s' from the Group buffer.  Revert to the old behaviour with
>>   `(setq gnus-agent nil)'.  Note that putting (gnus-agentize) in
>>   ~/.gnus is not needed any more.
>
> So, it is the case that nothing gets downloaded by default, even in
> agentized groups?.  That seems wrong.

Hm.  By default, in agentized groups, headers are downloaded into the
agent cache once and then retrieved from the agent cache only.
Regardless of pluggedness.

> By `Revert to the old behavior' do we mean revert to downloading
> agentized groups by default? Or, not having agent enabled by default?

Not having agent enabled by default.

> Depending on above answers:
> I don't see the advantage of not using gnus-agentize, but using 
> (setq gnus-agent nil) to do what `gnus-agentize' used to do.
> Especially since the syntax, on first glance, looks as if it would
> shut the agent off.

The first glance is correct.  To get the old behaviour, no agent at
all, you disable the agent, and in Oort you do this by setting
gnus-agent to nil.  Calling `gnus-agentize' is a obsolete way of
enabling the agent, equal to setting gnus-agent non-nil.

> Seems like asking for confusion.... but maybe its just me.

Is it less confusing now?  Is the text consistent with this?  Could
you perhaps suggest improved text?

> If a group is set agentized, it seems the default one would expect is
> that downloading of headers would be controlled by the default
> predicate.  Further categories would take it from there by user
> specifying them as always.

Categories only affect articles.  Header downloads are not affected by
predicates.

Re: Do not want agantized, nnml backend.

[Harry Putnam]

Simon Josefsson <jas@extundo.com> writes:

>> Seems like asking for confusion.... but maybe its just me.
>
> Is it less confusing now?  Is the text consistent with this?  Could
> you perhaps suggest improved text?

Yes, its less confusing after the explanantions.  So far as
suggesting improved text, it seems unlikely I could since I barely
understand it myself.

Sorry to be such a pain, but I think I finally `get it'.

Re: Do not want agantized, nnml backend.

[Jesper Harder]

Simon Josefsson <jas@extundo.com> writes:

> Another try:
>
>   The Gnus Agent is now enabled by default, [..]  Revert to the old
>   behaviour with `(setq gnus-agent nil)'.  Note that putting
>   (gnus-agentize) in ~/.gnus is not needed any more.

I think the node (gnus)Gnus Unplugged also needs to be updated.

Re: Do not want agantized, nnml backend.

[Simon Josefsson]

Jesper Harder <harder@ifa.au.dk> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> Another try:
>>
>>   The Gnus Agent is now enabled by default, [..]  Revert to the old
>>   behaviour with `(setq gnus-agent nil)'.  Note that putting
>>   (gnus-agentize) in ~/.gnus is not needed any more.
>
> I think the node (gnus)Gnus Unplugged also needs to be updated.

Right, I committed something.

Re: Do not want agantized, nnml backend.

[小関 吉則]

>>>>> In <ilulm6tbj9y.fsf@latte.josefsson.org> 
>>>>> Simon Josefsson <jas@extundo.com> wrote:
> > But when Unplugged, In Summary buffer marked %. I can't read
> > mails.
> >
> > Agentized backend is Only nntp and nnimap???

> Amazing, I was about to add a patch that agentizes nntp and nnimap by
> default, and it fixes your problem too.

> I've committed my patch now, let me know if it doesn't work.

Thank you, it works fine now!
-- 
こせき / アトム誕生まであと223日 @@)v