* Re: nnml pseudo-agentized?
2003-04-30 13:21 ` Kevin Greiner
@ 2003-04-30 16:32 ` Kai Großjohann
2003-04-30 21:42 ` Kai Großjohann
0 siblings, 1 reply; 5+ messages in thread
From: Kai Großjohann @ 2003-04-30 16:32 UTC (permalink / raw)
Kevin Greiner <kgreiner@xpediantsolutions.com> writes:
> I just fixed a bug in gnus-open-server. If you debug it, you should
> be able to see what is going wrong. I suspect that the new code
> doesn't understand that some backends are always plugged since they
> don't require network connections.
>
> In fact, I believe that the change should be at the '(not
> gnus-plugged)' condition. One fix would be to change it to
> "(not (or gnus-plugged (memq (car gnus-command-method) '(nnml))))".
> Seems simple except that I don't know which backends go into the list
> with nnml.
Your analysis is right. I applied the enclosed patch and then I
could successfully open the server.
However, I don't think that this is the right approach. I've been
thinking about it, and I think the key is that my nnml server isn't
agentized, so IMVHO gnus-open-server should open it anyway.
Consider somebody running a Leafnode or Hamster on their machine:
they would abstain from agentizing the server, and then Gnus should
connect to the server. Even though the server is nntp, and thus
requires network connectivity, and thus isn't mentioned in the list.
Hm. Let me change the condition to check gnus-agent-covered-methods
and see if it still works.
Yep. It seems to work. At least I could open the server.
Testing... Ah! I can hit `g', and enter an nnml group. So this
seems to be the right patch.
For reference, here is the patch that looks whether the method
requires network connectivity:
--- gnus-int.el.~6.30.~ Wed Apr 30 06:03:16 2003
+++ gnus-int.el Wed Apr 30 18:22:41 2003
@@ -207,16 +207,18 @@
((eq (nth 1 state) 'offline)
;; If this method was previously opened offline, we just return t.
t)
- ((not gnus-plugged)
+ ((and (not gnus-plugged)
+ (not (memq (car gnus-command-method)
+ '(nnml nnfolder nnmh))))
;; I'm opening servers while unplugged. Set the status to
;; either 'offline or 'denied without asking (I'm assuming
;; that the user wants to go 'offline on every agentized
;; server when opening while unplugged.)
(setcar (cdr state) (if (and gnus-agent
- (gnus-agent-method-p gnus-command-method))
- (or gnus-server-unopen-status
- 'offline)
- 'denied))
+ (gnus-agent-method-p gnus-command-method))
+ (or gnus-server-unopen-status
+ 'offline)
+ 'denied))
(if (eq (nth 1 state) 'offline)
;; Invoke the agent's backend to open the offline server.
And here is the patch that I'm currently using:
--- gnus-int.el.~6.30.~ Wed Apr 30 06:03:16 2003
+++ gnus-int.el Wed Apr 30 18:27:25 2003
@@ -207,16 +207,18 @@
((eq (nth 1 state) 'offline)
;; If this method was previously opened offline, we just return t.
t)
- ((not gnus-plugged)
+ ((and (not gnus-plugged)
+ (member gnus-command-method
+ gnus-agent-covered-methods))
;; I'm opening servers while unplugged. Set the status to
;; either 'offline or 'denied without asking (I'm assuming
;; that the user wants to go 'offline on every agentized
;; server when opening while unplugged.)
(setcar (cdr state) (if (and gnus-agent
- (gnus-agent-method-p gnus-command-method))
- (or gnus-server-unopen-status
- 'offline)
- 'denied))
+ (gnus-agent-method-p gnus-command-method))
+ (or gnus-server-unopen-status
+ 'offline)
+ 'denied))
(if (eq (nth 1 state) 'offline)
;; Invoke the agent's backend to open the offline server.
I propose to install the patch that I'm currently using. I will
commit it when I'm back online, but I'm sitting in the train at the
moment, and will go out tonight. It's 18:30 here (= 16:30 UTC) and I
don't expect that this will get sent before 22:00 (= 20:00 UTC). The
commit will have to wait for tomorrow, I'm afraid.
Thanks, Kevin, for pointing out what needs to be done. Gnus is so
cool after your Agent changes! I'm happy it works again.
--
file-error; Data: (Opening input file no such file or directory ~/.signature)
^ permalink raw reply [flat|nested] 5+ messages in thread