Re: disable "org-decrypt: auto-save-mode may cause leakage"

["Sebastien Vauban" <wxhgmqzgwmuf-geNee64TY+gS+FvcfC7Uqw@public.gmane.org>]

Hi Pieter,

Pieter Praet wrote:
> On Thu, 14 Jul 2011 16:42:58 +0200, "Sebastien Vauban" <wxhgmqzgwmuf-geNee64TY+gS+FvcfC7Uqw@public.gmane.org> wrote:
>> Pieter Praet wrote:
>>> On Thu, 14 Jul 2011 15:37:34 +0200, "Sebastien Vauban" <wxhgmqzgwmuf-geNee64TY+gS+FvcfC7Uqw@public.gmane.org> wrote:
>>>> Shouldn't its default value better be set to 'encrypt?
>>>
>>> I don't think so.
>>>
>>> It only re-encrypts entries which are tagged :crypt: (or whatever you
>>> have `org-crypt-tag-matcher' set to), so I'd prefer users to be "forced"
>>> to at least read the docstring so as not to be using it with a false
>>> sense of security.
>>>
>>> Though this could be easily solved by automatically tagging entries upon
>>> encryption, I think it's more "considerate" to simply (ask to) disable
>>> `auto-save-mode' for the current buffer, as it's quite annoying to have
>>> an entry get re-encrypted while you're in the middle of typing something.
>>
>> This is more clear to me, and makes a lot of sense.
>>
>> Though, what I find annoying, and the OP as well, is to be asked that
>> question for buffers which do not contain any encrypted entry. Would it be
>> possible to scan the buffer for such entries, and only ask the user when
>> the buffer is potentially exposed?
>
> AFAIK, that should already be the case.
>
> Even when running with the following config...
>
>   #+BEGIN_SRC emacs-lisp
>     (add-hook 'find-file-hook 'org-decrypt-entries)
>   #+END_SRC
>
> ... the user shouldn't get prompted needlessly, because
> `org-decrypt-entries' applies `org-decrypt-entry' (which does the
> `auto-save-mode' check) *only* to entries with a tag matching
> `org-crypt-tag-matcher' [1].
>
> So prompting should only occur for buffers for which `auto-save-mode' is
> enabled AND which contain one or more entries tagged with `crypt'
> (regardless of whether the entry is actually encrypted, but that's beside
> the point) UNLESS running `org-decrypt-entry' (auto-)manually for each
> buffer.
>
> I suspect it's a user config issue.

The "problem" I encounter is difficult to narrow, but I can tell you now that,
in my case, the question is asked for buffers which do not contain any tagged
entry with "crypt" (neither the string "crypt" as word or subword of that
buffer):

I simply call C-c C-x C-j to jump to the current clocking entry, and the
question "org-decrypt: auto-save-mode may cause leakage. Disable it for
current buffer? (y or n)" pops up -- while the current buffer has nothing
special, as said above.

This is reproducible for me, with Git version of this morning.

Best regards,
  Seb

-- 
Sebastien Vauban