connect timeouts?

connect timeouts?

[robert]

Hi, to what extend are timeouts used in connect in gnus? I figure it's
connect() that's doing all the work in Emacs ..

Anyway, I'm getting a lot of times where I'll "space" on a foreign group in
the group buffer and it will just say cannot open server. 

I'll telnet to my hosts news server and it will eventually come back with
a 200

-- 
You are amazed that they exist and they burn so 
bright whilst you can only wonder why.

Re: connect timeouts?

[Lars Magne Ingebrigtsen]

robert@elastica.com writes:

> Hi, to what extend are timeouts used in connect in gnus? I figure it's
> connect() that's doing all the work in Emacs ..

By default there are no timeouts in Gnus.  (Well, unless there are
some Emacs timeouts that takes place.)

> Anyway, I'm getting a lot of times where I'll "space" on a foreign
> group in the group buffer and it will just say cannot open server.

That means that connection has previously been denied, so Gnus refuses
to try again.  `^' in the group buffer to see the status of the
servers.

-- 
  "Yes.  The journey through the human heart 
     would have to wait until some other time."

Re: connect timeouts?

[Michael Lamoureux]

 "lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

lars> That means that connection has previously been denied, so Gnus
lars> refuses to try again.

Why?  Our news server gets flakey every once in a while, and if I try
to connect at the wrong time, then gnus is basically useless for
reading news until I restart (I can, of course, still read mail, but
that's not the point).  This is silly, because as soon as the news
server is rebooted, it's ok, but how am I supposed to know when that
is if gnus keeps refusing to connect?  (there are actually a handful
of ways, but none of them are as easy as hitting 'g' ;-)

I figure if I'm stupid enough to keep trying to connecting to a server
that I'm not allowed to access, you should let me beat my head against
the wall until it bleeds...


IMHO,
Michael

Re: connect timeouts?

[Lars Magne Ingebrigtsen]

Michael Lamoureux <lamour@engin.umich.edu> writes:

> Why?  Our news server gets flakey every once in a while, and if I try
> to connect at the wrong time, then gnus is basically useless for
> reading news until I restart (I can, of course, still read mail, but
> that's not the point).  This is silly, because as soon as the news
> server is rebooted, it's ok, but how am I supposed to know when that
> is if gnus keeps refusing to connect?  (there are actually a handful
> of ways, but none of them are as easy as hitting 'g' ;-)
> 
> I figure if I'm stupid enough to keep trying to connecting to a server
> that I'm not allowed to access, you should let me beat my head against
> the wall until it bleeds...

Well, yes.  But no.  :-) Say you read ten foreign groups from the
"nntp:flaky" server, and you press `g'.  Gnus will then try to
activate these ten groups.  If the first attempt at connecting fails,
it's probably not very productive to try to reconnect to the server 9
more times?

Perhaps there should be a time-out to the denials, but I don't think
so.  Just use `^' to see the server status and remove the denials at
will. 

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@ifi.uio.no * Lars Ingebrigtsen

Re: connect timeouts?

[Michael Lamoureux]

 "lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

lars> Michael Lamoureux <lamour@engin.umich.edu> writes:
>> I figure if I'm stupid enough to keep trying to connecting to a
>> server that I'm not allowed to access, you should let me beat my
>> head against the wall until it bleeds...

lars> Well, yes.  But no.  :-) Say you read ten foreign groups from
lars> the "nntp:flaky" server, and you press `g'.  Gnus will then try
lars> to activate these ten groups.  If the first attempt at
lars> connecting fails, it's probably not very productive to try to
lars> reconnect to the server 9 more times?

No, but mine is not a foreign server...  Does gnus not keep track of
which is the primary server and which are foreign?


lars> Perhaps there should be a time-out to the denials, but I don't
lars> think so.

perhaps.  Or maybe it should just skip that server for the remainder
of *this* pass (i.e., until the 'g' operation has completed).  That
way I get the benefit that you mention as well as the operation that I
want.


lars> Just use `^' to see the server status and remove the denials at
lars> will.

I'll take a look at this...


thanks,
Michael

Re: connect timeouts?

[Lars Magne Ingebrigtsen]

Michael Lamoureux <lamour@engin.umich.edu> writes:

> lars> Perhaps there should be a time-out to the denials, but I don't
> lars> think so.
> 
> perhaps.  Or maybe it should just skip that server for the remainder
> of *this* pass (i.e., until the 'g' operation has completed).  That
> way I get the benefit that you mention as well as the operation that I
> want.

Well, I would find it supremely annoying if Gnus were to try to
reconnect to a server that's down each time I press `g'.  (I might
just want to re-scan my mail groups, for instance.)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@ifi.uio.no * Lars Ingebrigtsen

Re: connect timeouts?

[Michael Welsh Duggan]

Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

> 
> Michael Lamoureux <lamour@engin.umich.edu> writes:
> 
> > lars> Perhaps there should be a time-out to the denials, but I don't
> > lars> think so.
> > 
> > perhaps.  Or maybe it should just skip that server for the remainder
> > of *this* pass (i.e., until the 'g' operation has completed).  That
> > way I get the benefit that you mention as well as the operation that I
> > want.
> 
> Well, I would find it supremely annoying if Gnus were to try to
> reconnect to a server that's down each time I press `g'.  (I might
> just want to re-scan my mail groups, for instance.)

I agree.  However, I think that this _should_ be the default behavior
when one presses M-g on a inoperative group.

-- 
Michael Duggan
(md5i@schenley.com)

Re: connect timeouts?

[Lars Magne Ingebrigtsen]

"Michael Welsh Duggan" <md5i@schenley.com> writes:

> However, I think that this _should_ be the default behavior when one
> presses M-g on a inoperative group.

Yup.  I've added this to the Red Gnus todo list.

-- 
  "Yes.  The journey through the human heart 
     would have to wait until some other time."