nnimap now tries both imaps and imap

nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

I finally remembered to implement the thing previously discussed where
nnimap first tries imaps, and if it can't get a connection there, then
it tries imap.  This is unless the user has set `nnimap-stream' or
`nnimap-server-port' explicitly, of course.

It doesn't save what it discovers, so it'll do this probing each time it
connects.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

Lars wrote:

> ... nnimap first tries imaps, and if it can't get a connection there, then
> it tries imap....

Is something similar happening with nntp? I've just started using the
git version (instead of 5.13) and it's attempting to use STARTTLS to
get to news.gmane.org (and hanging in the process until I hit
C-g). Relevant setup is just a simple:

(setq gnus-select-method '(nntp "news.gmane.org"))

Changing to the following had no effect:

(setq gnus-select-method '(nntp "news.gmane.org"
		             (nntp-open-connection-function nntp-open-network-stream)
		             ))

Tommy

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

> Is something similar happening with nntp? I've just started using the
> git version (instead of 5.13) and it's attempting to use STARTTLS to
> get to news.gmane.org (and hanging in the process until I hit
> C-g). Relevant setup is just a simple:
>
> (setq gnus-select-method '(nntp "news.gmane.org"))

Hm.  Interesting.  What do the following two forms eval to?

(fboundp 'open-gnutls-stream)

(executable-find "gnutls-cli")

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

> Hm.  Interesting.  What do the following two forms eval to?
>
> (fboundp 'open-gnutls-stream)

nil

> (executable-find "gnutls-cli")

"/opt/local/bin/gnutls-cli"

Tommy

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

>> (executable-find "gnutls-cli")
>
> "/opt/local/bin/gnutls-cli"

Right, so it's trying to use gnutls-cli, most likely.  Could you try to
edebug through `proto-stream-open-network' and see where it hangs?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

> Could you try to
> edebug through `proto-stream-open-network' and see where it hangs?

Phew. I can try - but I don't know lisp and have only just heard of
edebug. I may take a while!

t

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

> Phew. I can try - but I don't know lisp and have only just heard of
> edebug. I may take a while!

I've made some more changes to the code (basically disabling it by
default if you don't have built-in gnutls support), so it should
probably work again for you now.

But it would still be interesting to know why it failed...

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

> But it would still be interesting to know why it failed...

I am clueless about this stuff, but diving in nevertheless ... (warning:
you're going to end up more pushing me past newbie edebug questions than
getting anything sensible on the debug front):

So, I found the defun you mentioned, loaded edebug (I think), and then
ran eval-defun with an argument on it. I'm led to believe I have now
instrumented the function so as to get a breakpoint. (Remember -- I know
nothing).

I then run gnus and execution stops at the function. I then step through
(using "space", right?). It goes all the way to loading my secondary
select methods -- i.e. it does not hang. However, it doesn't actually
get nntp going, evidenced by no new messages in the relevant groups
(despite me knowing there are some there).

However, if instead of stepping I hit "G' (or "C' etc), to let it start
running again, it hangs as it tries nntp.

That's all I know (if that) for now.

t

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

> I then run gnus and execution stops at the function. I then step through
> (using "space", right?). It goes all the way to loading my secondary
> select methods -- i.e. it does not hang. 

Ah, ok.  Then is sounds quite a lot like the race condition I fixed half
an hour ago.  It would usually work for me, and always when I stepped
though, but sometimes it would fail.  This should now hopefully be fixed.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> I've made some more changes to the code (basically disabling it by
> default if you don't have built-in gnutls support), so it should
> probably work again for you now.

And it did. Thanks!

My newbness knows no bounds, but if you could tell me how in git to
quickly compare what you now have with what you had before, I can keep
banging my head off it to see if I can give you a clue as to what the
issue is.

t

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

> My newbness knows no bounds, but if you could tell me how in git to
> quickly compare what you now have with what you had before, I can keep
> banging my head off it to see if I can give you a clue as to what the
> issue is.

Using git is still a mystery to me.  :-)

But if you

(setq proto-stream-always-use-starttls t)

you'll re-enable the STARTTLS logic.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

> Ah, ok.  Then is sounds quite a lot like the race condition ...

Ooer. Cancel that previous request for git assistance on diffing the
change. Cutting my lisp debugging teeth on a race condition sounds like
a path to madness :-)

Tommy

Re: nnimap now tries both imaps and imap

[Tommy Kelly]

> But if you
>
> (setq proto-stream-always-use-starttls t)
>
> you'll re-enable the STARTTLS logic.

Is that saying "always use starttls if the server reports being able to
handle it"? In which case, was my problem that gmane appeared to be able
to handle it but actually can't?

Or is the above saying "always use starttls unless I tell you otherwise
(e.g. in the select method using nntp-open-connection-function)"? In
which case, what's the correct mechanism for turning off starttls?  

Tommy

Re: nnimap now tries both imaps and imap

[Andreas Schwab]

Tommy Kelly <tommy.kelly@verilab.com> writes:

> My newbness knows no bounds, but if you could tell me how in git to
> quickly compare what you now have with what you had before, I can keep
> banging my head off it to see if I can give you a clue as to what the
> issue is.

$ git diff HEAD@{1}..

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Tommy Kelly <tommy.kelly@verilab.com> writes:

>> But if you
>>
>> (setq proto-stream-always-use-starttls t)
>>
>> you'll re-enable the STARTTLS logic.
>
> Is that saying "always use starttls if the server reports being able to
> handle it"? In which case, was my problem that gmane appeared to be able
> to handle it but actually can't?

Yes.

It defaults to t when you have built-in gnutls support, but not if
you're using gnutls-cli.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nnimap now tries both imaps and imap

[Steinar Bang]

>>>>> Lars Magne Ingebrigtsen <larsi@gnus.org>:

> Tommy Kelly <tommy.kelly@verilab.com> writes:
>> My newbness knows no bounds, but if you could tell me how in git to
>> quickly compare what you now have with what you had before, I can keep
>> banging my head off it to see if I can give you a clue as to what the
>> issue is.

> Using git is still a mystery to me.  :-)

Well, er... 
 - open the file in emacs
 - Do `C-u C-x v =', and give the guid of the second-to-last commit as
   the version to diff from, and current version as the version

Of course, this was a bit easier in version control systems that had a
human-readable version number.

Alternatively you can:
 - open the file in emacs
 - Do `C-x v l'
 - If you end up the end of the log (which happens to me on git for some
   reason), arrow up to the second to last entry (ie. navigate in a way
   that doesn't set mark) and press `d'

...or you can pick out the guid for the commit you'd like to diff
against and do the `C-u C-x v =' thing.

Re: nnimap now tries both imaps and imap

[Adam Sjøgren]

On Mon, 29 Nov 2010 09:56:08 +0100, Steinar wrote:

> ...or you can pick out the guid for the commit you'd like to diff
> against and do the `C-u C-x v =' thing.

When you say "guid", I guess you mean sha1 hash?


  Best regards,

    Adam

-- 
 "You have to photosynthesize"                                Adam Sjøgren
                                                         asjo@koldfront.dk

Re: nnimap now tries both imaps and imap

[Steinar Bang]

>>>>> asjo@koldfront.dk (Adam Sjøgren):

> When you say "guid", I guess you mean sha1 hash?

Whatever.

Re: nnimap now tries both imaps and imap

[Lars Magne Ingebrigtsen]

Steinar Bang <sb@dod.no> writes:

> Well, er... 
>  - open the file in emacs
>  - Do `C-u C-x v =', and give the guid of the second-to-last commit as
>    the version to diff from, and current version as the version
>
> Of course, this was a bit easier in version control systems that had a
> human-readable version number.

A bit.  :-)

> Alternatively you can:
>  - open the file in emacs
>  - Do `C-x v l'
>  - If you end up the end of the log (which happens to me on git for some
>    reason), arrow up to the second to last entry (ie. navigate in a way
>    that doesn't set mark) and press `d'
>
> ...or you can pick out the guid for the commit you'd like to diff
> against and do the `C-u C-x v =' thing.

Yeah, that's a nice way of getting the diff between two versions.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen