gmail and pop3

gmail and pop3

[Alex Romosan]

after not being able to log on my gmail account using gnus and pop3 for
the last couple of months i finally decided to do a git bisect to find
the commit that broke it and came up with this:

152b8331ec8e0d49c005b2b19bfacc77df882c46 is the first bad commit
commit 152b8331ec8e0d49c005b2b19bfacc77df882c46
Author: John Wiegley <jwiegley@gmail.com>
Date:   Tue Dec 25 18:31:47 2012 +0100

    Allow "password" as netrc passwords
    
    * auth-source.el (auth-source-netrc-parse): Allow using "password" as
    the password (bug#12097).

:040000 040000 2cf066be80165460b8e824414098c87f22000089 548900db7ba3abbefe3c201fd304892dcf398c2c M      lisp


furthermore, if i rename my .netrc file to something else then i can
access my gmail account. so it looks like this patch rather than allow
"password" as netrc passwords made it as require password if .netrc
exists.

--alex--

-- 
| I believe the moment is at hand when, by a paranoiac and active |
|  advance of the mind, it will be possible (simultaneously with  |
|  automatism and other passive states) to systematize confusion  |
|  and thus to help to discredit completely the world of reality. |

Re: gmail and pop3

[John Wiegley]

>>>>> Alex Romosan <romosan@sycorax.lbl.gov> writes:

> furthermore, if i rename my .netrc file to something else then i can access
> my gmail account. so it looks like this patch rather than allow "password"
> as netrc passwords made it as require password if .netrc exists.

Sorry about, Alex.  I did find the parsing code in this area of Gnus rather
difficult to reason about.  Maybe someone with more familiarity would be
willing to look over my commit?

Thanks,
  John

Re: gmail and pop3

[Alex Romosan]

"John Wiegley" <jwiegley@gmail.com> writes:

>>>>>> Alex Romosan <romosan@sycorax.lbl.gov> writes:
>
>> furthermore, if i rename my .netrc file to something else then i can
>> access my gmail account. so it looks like this patch rather than
>> allow "password" as netrc passwords made it as require password if
>> .netrc exists.
>
> Sorry about, Alex. I did find the parsing code in this area of Gnus
> rather difficult to reason about. Maybe someone with more familiarity
> would be willing to look over my commit?

no problem. hopefully somebody will find the solution so i can log on
gmail and keep my .netrc file. or, i guess, i can just try to add my
gmail username/password to the .netrc file (which i haven't tried yet).

--alex--

-- 
| I believe the moment is at hand when, by a paranoiac and active |
|  advance of the mind, it will be possible (simultaneously with  |
|  automatism and other passive states) to systematize confusion  |
|  and thus to help to discredit completely the world of reality. |

Re: gmail and pop3

[Ted Zlatanov]

On Fri, 01 Mar 2013 14:33:50 -0600 "John Wiegley" <jwiegley@gmail.com> wrote: 

>>>>>> Alex Romosan <romosan@sycorax.lbl.gov> writes:
>> furthermore, if i rename my .netrc file to something else then i can access
>> my gmail account. so it looks like this patch rather than allow "password"
>> as netrc passwords made it as require password if .netrc exists.

JW> Sorry about, Alex.  I did find the parsing code in this area of Gnus rather
JW> difficult to reason about.

It's a bit outdated and can't handle multi-line netrc entries, so I'll
probably rewrite it at some point.

JW> Maybe someone with more familiarity would be willing to look over my
JW> commit?

I have no idea what your commit:

+                   ((and (member elem tokens) (null pair))

does without re-reading all of the surrounding code.  Can you give an
example of what it supports that didn't work before?  I thought quoted
password were already supported.

Thanks!
Ted

Re: gmail and pop3

[Ted Zlatanov]

On Thu, 06 Jun 2013 11:43:51 -0400 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> On Fri, 01 Mar 2013 14:33:50 -0600 "John Wiegley" <jwiegley@gmail.com> wrote: 
>>>>>>> Alex Romosan <romosan@sycorax.lbl.gov> writes:
>>> furthermore, if i rename my .netrc file to something else then i can access
>>> my gmail account. so it looks like this patch rather than allow "password"
>>> as netrc passwords made it as require password if .netrc exists.

JW> Sorry about, Alex.  I did find the parsing code in this area of Gnus rather
JW> difficult to reason about.

TZ> It's a bit outdated and can't handle multi-line netrc entries, so I'll
TZ> probably rewrite it at some point.

JW> Maybe someone with more familiarity would be willing to look over my
JW> commit?

TZ> I have no idea what your commit:

TZ> +                   ((and (member elem tokens) (null pair))

TZ> does without re-reading all of the surrounding code.  Can you give an
TZ> example of what it supports that didn't work before?  I thought quoted
TZ> password were already supported.

After sending this, I saw bug#12097 which has some background on this.

So definitely a TODO for me is to rewrite that parser.  Sorry for the
trouble.

Ted

netrc parser revamp (was: gmail and pop3)

[Ted Zlatanov]

On Thu, 06 Jun 2013 14:06:35 -0400 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> After sending this, I saw bug#12097 which has some background on this.

TZ> So definitely a TODO for me is to rewrite that parser.  Sorry for the
TZ> trouble.

Take a look at the following.  It implements a parser that will go
through the current buffer and extract all the netrc entries, respecting
multiline entries and other weird things.  Also, netrc used single
quotes (double quotes are a Gnusism apparently), so this supports both
single and double quotes.  Entries must start with "machine" which is
required in the netrc format.

It doesn't treat "macdef" entries especially because I have never seen
them used by anyone and didn't want the extra complexity.  If you feel
they must be supported, speak up.

e.g. try it on this netrc content:

#+begin_src text
machine A login tzz@lifelogs.com password "A lovely A"
# test again
machine B login tzz@lifelogs.com
 password 'B single quotes B'
default other info goes here
#+end_src

If this works as expected I'll put it into auth-source.el.  So please
try it on your weird wacky netrc files :)

Thanks
Ted

#+begin_src lisp
(defun auth-source-netrc-parse-next-interesting ()
  "Advance to the next interesting position in the current buffer."
  ;; If we're looking at a comment or are at the end of the line, move forward
  (while (or (looking-at "#")
             (and (eolp)
                  (not (eobp))))
    (forward-line 1))
  (skip-chars-forward "\t "))

(defun auth-source-netrc-parse-one ()
  "Read one thing from the current buffer."
  (auth-source-netrc-parse-next-interesting)

  (when (or (looking-at "'\\([^']+\\)'")
            (looking-at "\"\\([^\"]+\\)\"")
            (looking-at "\\([^ \t\n]+\\)"))
    (forward-char (length (match-string 0)))
    (auth-source-netrc-parse-next-interesting)
    (match-string-no-properties 1)))

(defun auth-source-netrc-parse-entries ()
  "Parse all the netrc entries from the current buffer."
  (let (item item2 all alist default)
    (while (setq item (auth-source-netrc-parse-one))

      (setq default (equal item "default"))
      ;; We're starting a new machine.  Save the old one.
      (when (or default
                (equal item "machine"))
        (when alist
          (push alist all))
        (setq alist nil))

      ;; In default entries, we don't have a next token.
      ;; We store them as ("machine" . t)
      (if default
          (push (cons "machine" t) alist)
        ;; Not a default entry.  Grab the next item.
        (when (setq item2 (auth-source-netrc-parse-one))
          (push (cons item item2) alist))))
    (when alist
      (push alist all))
    (nreverse all)))

#+end_src

Re: netrc parser revamp

[Ted Zlatanov]

On Mon, 10 Jun 2013 02:26:42 -0400 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> On Thu, 06 Jun 2013 14:06:35 -0400 Ted Zlatanov <tzz@lifelogs.com> wrote: 
TZ> After sending this, I saw bug#12097 which has some background on this.

TZ> So definitely a TODO for me is to rewrite that parser.  Sorry for the
TZ> trouble.

I've comitted a new version of the netrc parser that can handle
single-quoted strings and multiline entries.  I've tested it but please
let me know if you notice issues.

Ted