bug#27445: 26.0.50; Insufficient documentation for pinentry.el

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Michael Heerdegen]

Hallo,

I followed the instructions in pinentry.el, but failed.  There is no
more documentation available.  I'm running a very normal Debian.

Since entering passwords is a quite common thing, and this is about the
user's security, shouldn't this deserve some more extensive information
or be described in the manual?

FWIW, after searching the Internet, I found several pieces of
information that seem to be missing in the documentation in the file
header of pinentry.el: I seemingly have to build pinentry-emacs if it's
not available, for example.

After having tried over an hour, I finally was in a state where I got no
more password prompts at all.

I'm thankful that I had the luck to be able to restore the previous
state so that I got my Email working again and could open this report.
This should not happen to others.


Thanks in advance,

Michael.


In GNU Emacs 26.0.50 (build 3, x86_64-pc-linux-gnu, GTK+ Version 3.22.11)
 of 2017-06-19 built on drachen
Repository revision: 1b75af59b305867c89271905be72a05d06a4eff4
Windowing system distributor 'The X.Org Foundation', version 11.0.11903000
System Description:	Debian GNU/Linux 9.0 (stretch)

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Michael Heerdegen]

Michael Heerdegen <michael_heerdegen@web.de> writes:

> I'm thankful that I had the luck to be able to restore the previous
> state so that I got my Email working again and could open this report.
> This should not happen to others.

I discovered by accident that

#+begin_src emacs-lisp
  (setq epa-pinentry-mode 'loopback)
#+end_src

does what I want.

It would still be good to have some page in the manual that describes
with what tools file encryption in Emacs is done, and where to look for
documentation.  Currently there is only a description of encryption
commands of dired (info "(emacs) Operating on Files").


Michael.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Michael Heerdegen <michael_heerdegen@web.de>
> Date: Thu, 26 Oct 2017 14:00:27 +0200
> 
> Michael Heerdegen <michael_heerdegen@web.de> writes:
> 
> > I'm thankful that I had the luck to be able to restore the previous
> > state so that I got my Email working again and could open this report.
> > This should not happen to others.
> 
> I discovered by accident that
> 
> #+begin_src emacs-lisp
>   (setq epa-pinentry-mode 'loopback)
> #+end_src
> 
> does what I want.
> 
> It would still be good to have some page in the manual that describes
> with what tools file encryption in Emacs is done, and where to look for
> documentation.  Currently there is only a description of encryption
> commands of dired (info "(emacs) Operating on Files").

I agree.

Daiki, could you please add some user-level documentation about this
package?

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

Ping!

> Date: Sat, 28 Oct 2017 12:13:20 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 27445@debbugs.gnu.org
> 
> > From: Michael Heerdegen <michael_heerdegen@web.de>
> > Date: Thu, 26 Oct 2017 14:00:27 +0200
> > 
> > Michael Heerdegen <michael_heerdegen@web.de> writes:
> > 
> > > I'm thankful that I had the luck to be able to restore the previous
> > > state so that I got my Email working again and could open this report.
> > > This should not happen to others.
> > 
> > I discovered by accident that
> > 
> > #+begin_src emacs-lisp
> >   (setq epa-pinentry-mode 'loopback)
> > #+end_src
> > 
> > does what I want.
> > 
> > It would still be good to have some page in the manual that describes
> > with what tools file encryption in Emacs is done, and where to look for
> > documentation.  Currently there is only a description of encryption
> > commands of dired (info "(emacs) Operating on Files").
> 
> I agree.
> 
> Daiki, could you please add some user-level documentation about this
> package?

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Eli Zaretskii <eliz@gnu.org> writes:

> Ping!

Sorry for not responding to it earlier.

>> Daiki, could you please add some user-level documentation about this
>> package?

I would rather not document it, because the package should be
transparent to user.  If it doesn't work, you are using either GnuPG 2.0
or some stupid distro package neglecting upstream's default (only
Debian, as far as I know).

Anyway; the package (and the corresponding change in GnuPG and pinentry)
was intended to provide a way to input passphrase through Emacs with
GnuPG 2.0.  However for some reason the change was only implemented in
GnuPG >= 2.1 and didn't get backported to GnuPG 2.0.

>> > I discovered by accident that
>> > 
>> > #+begin_src emacs-lisp
>> >   (setq epa-pinentry-mode 'loopback)
>> > #+end_src
>> > 
>> > does what I want.

This is indeed a correct solution if we only care about GnuPG >= 2.1.

Fortunately, GnuPG 2.0 will be unsupported in the next month.  After
that we should be able to obsolete or remove the package.

Regards,
-- 
Daiki Ueno

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Andy Moreton]

On Sun 19 Nov 2017, Daiki Ueno wrote:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> Ping!
>
> Sorry for not responding to it earlier.
>
>>> Daiki, could you please add some user-level documentation about this
>>> package?
>
> I would rather not document it, because the package should be
> transparent to user.  If it doesn't work, you are using either GnuPG 2.0
> or some stupid distro package neglecting upstream's default (only
> Debian, as far as I know).

How on earth is a user meant to diagnose a non-working setup without
adequate documentation ?

> Fortunately, GnuPG 2.0 will be unsupported in the next month.  After
> that we should be able to obsolete or remove the package.

Just because upstream remove support for something does not mean that
all downstream distros or users systems are fully up to date.

While it is sensible to encourage users to use the latest package
version and to install security fixes, there will be many people with
systems using older package versions for some time. Those users will
need documentation to guide them.

    AndyM

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Michael Heerdegen]

Daiki Ueno <ueno@gnu.org> writes:

> I would rather not document it, because the package should be
> transparent to user.  If it doesn't work, you are using either GnuPG 2.0
> or some stupid distro package neglecting upstream's default (only
> Debian, as far as I know).
>
> Anyway; the package (and the corresponding change in GnuPG and pinentry)
> was intended to provide a way to input passphrase through Emacs with
> GnuPG 2.0.  However for some reason the change was only implemented in
> GnuPG >= 2.1 and didn't get backported to GnuPG 2.0.

I don't fully understand the backgrounds.  Does what you say mean that
the problem will go away very soon (even on Debian) from alone?


Thanks,

Michael.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Michael Heerdegen <michael_heerdegen@web.de> writes:

> I don't fully understand the backgrounds.  Does what you say mean that
> the problem will go away very soon (even on Debian) from alone?

As I said in the previous post:

> Fortunately, GnuPG 2.0 will be unsupported in the next month.  After
> that we should be able to obsolete or remove the package.

As long as we support only GnuPG >= 2.1 and GnuPG 1.x, there is a way to
intercept passphrase prompt.  In that case, there will be no need to use
a kludge like pinentry.el (+ emacs support in GnuPG/pinentry).

Regards,
-- 
Daiki Ueno

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Andy Moreton <andrewjmoreton@gmail.com> writes:

> How on earth is a user meant to diagnose a non-working setup without
> adequate documentation ?

Why on earth the user could even notice the intentionally undocumented
package, then?

>> Fortunately, GnuPG 2.0 will be unsupported in the next month.  After
>> that we should be able to obsolete or remove the package.
>
> Just because upstream remove support for something does not mean that
> all downstream distros or users systems are fully up to date.
>
> While it is sensible to encourage users to use the latest package
> version and to install security fixes, there will be many people with
> systems using older package versions for some time. Those users will
> need documentation to guide them.

That's the task of the distros which provide extended life support for
GnuPG 2.0, not of the Emacs upstream.  Given that both Emacs and GnuPG
are part of the GNU project, they should be in sync in upstream.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Daiki Ueno <ueno@gnu.org>
> Date: Tue, 21 Nov 2017 13:15:13 +0100
> Cc: 27445@debbugs.gnu.org
> 
> Andy Moreton <andrewjmoreton@gmail.com> writes:
> 
> > How on earth is a user meant to diagnose a non-working setup without
> > adequate documentation ?
> 
> Why on earth the user could even notice the intentionally undocumented
> package, then?

I don't think we can claim that it's undocumented: it has some user
documentation in the Commentary section.

My take from this discussion is that this package will become
obsolete, but not exactly tomorrow.  So I think it would be good to
add some minimal verbiage to allow those who still need pinentry.el to
use it.  I gather the required additions are not too large.

Thanks.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Eli Zaretskii <eliz@gnu.org> writes:

> I don't think we can claim that it's undocumented: it has some user
> documentation in the Commentary section.
>
> My take from this discussion is that this package will become
> obsolete, but not exactly tomorrow.

Not tomorrow, but it will be just a month later.  Also note that this is
a relatively new package and no Emacs release has shipped with it.

> So I think it would be good to add some minimal verbiage to allow
> those who still need pinentry.el to use it.  I gather the required
> additions are not too large.

Sorry, I have no idea what to add to the existing documentation (the
Commentary section), really.  As I indicated in the previous post, if
pinentry.el doesn't work, it is likely to be failing in somewhere else
than Emacs (e.g. Debian packaging of GnuPG and pinentry).  It's not
obvious to me what we should suggest to the user in that case.

Regards,
-- 
Daiki Ueno

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Daiki Ueno <ueno@gnu.org>
> Cc: 27445@debbugs.gnu.org
> Date: Tue, 21 Nov 2017 20:59:19 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I don't think we can claim that it's undocumented: it has some user
> > documentation in the Commentary section.
> >
> > My take from this discussion is that this package will become
> > obsolete, but not exactly tomorrow.
> 
> Not tomorrow, but it will be just a month later.  Also note that this is
> a relatively new package and no Emacs release has shipped with it.

Then maybe we should simply remove it now?

However, what Michael wrote and your response to that confuse me now,
see below.

> > So I think it would be good to add some minimal verbiage to allow
> > those who still need pinentry.el to use it.  I gather the required
> > additions are not too large.
> 
> Sorry, I have no idea what to add to the existing documentation (the
> Commentary section), really.  As I indicated in the previous post, if
> pinentry.el doesn't work, it is likely to be failing in somewhere else
> than Emacs (e.g. Debian packaging of GnuPG and pinentry).  It's not
> obvious to me what we should suggest to the user in that case.

Previously, you wrote:

> Anyway; the package (and the corresponding change in GnuPG and pinentry)
> was intended to provide a way to input passphrase through Emacs with
> GnuPG 2.0.  However for some reason the change was only implemented in
> GnuPG >= 2.1 and didn't get backported to GnuPG 2.0.
> 
> >> > I discovered by accident that
> >> > 
> >> > #+begin_src emacs-lisp
> >> >   (setq epa-pinentry-mode 'loopback)
> >> > #+end_src
> >> > 
> >> > does what I want.
> 
> This is indeed a correct solution if we only care about GnuPG >= 2.1.
> 
> Fortunately, GnuPG 2.0 will be unsupported in the next month.  After
> that we should be able to obsolete or remove the package.

So now I'm confused.  Can you answer the following questions, and
thus help me see the light?

  . is pinentry.el needed when using GnuPG 2.1?
  . if not, why did you say that the above setting of
    epa-pinentry-mode is the correct solution for GnuPG 2.1?
  . how should Emacs users input passphrases with GnuPG 2.1, without
    using pinentry.el?

Thanks.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Eli Zaretskii <eliz@gnu.org> writes:

> Then maybe we should simply remove it now?

Let's do that.

> So now I'm confused.  Can you answer the following questions, and
> thus help me see the light?
>
>   . is pinentry.el needed when using GnuPG 2.1?

No.

>   . if not, why did you say that the above setting of
>     epa-pinentry-mode is the correct solution for GnuPG 2.1?

epa-pinentry-mode is orthogonal to pinentry.el.  What I'm saying is that
we shouldn't expand the documentation of pinentry.el, because it's
turning out to be a failed attempt.

>   . how should Emacs users input passphrases with GnuPG 2.1, without
>     using pinentry.el?

Through the graphical prompt, or using the epa-pinentry-mode 'loopback'
setting.

Regards,
-- 
Daiki Ueno

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Michael Heerdegen]

Daiki Ueno <ueno@gnu.org> writes:

> Through the graphical prompt, or using the epa-pinentry-mode 'loopback'
> setting.

When a user doesn't want a graphical prompt (I don't, because it freezes
Emacs, and I use Emacs to get my password), how can he know that he must
configure `epa-pinentry-mode'?  I only found it by making extensive use
of the debugger.  I didn't even know where to look and which libraries
were relevant.


Michael.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Michael Heerdegen <michael_heerdegen@web.de> writes:

>> Through the graphical prompt, or using the epa-pinentry-mode 'loopback'
>> setting.
>
> When a user doesn't want a graphical prompt (I don't, because it freezes
> Emacs, and I use Emacs to get my password), how can he know that he must
> configure `epa-pinentry-mode'?  I only found it by making extensive use
> of the debugger.  I didn't even know where to look and which libraries
> were relevant.

Note that it's practically impossible to make it seamlessly work; there
is no way to reliably detect that the 'loopback' pinentry is enabled by
the GnuPG installation.  I don't want to document non-working setup.

Use it at your own risk or blame someone else (the GnuPG upsteam or
downstream package maintainers), really.  I am almost done with this
issue.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Michael Heerdegen]

Daiki Ueno <ueno@gnu.org> writes:

> Use it at your own risk or blame someone else
> [...]
> I am almost done with this issue.

If you say it makes no sense to document that stuff, I can't argue
against that.  Sorry if you feel blamed (actually I didn't blame anyone)
- but if this is to be "used at own risk" and not worth or possible to
be documented decently, why has it been added to Emacs?  Isn't it normal
that people use it then and are left with questions when problems occur
and there is no documentation?


Michael.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Michael Heerdegen <michael_heerdegen@web.de> writes:

> Daiki Ueno <ueno@gnu.org> writes:
>
>> Use it at your own risk or blame someone else
>> [...]
>> I am almost done with this issue.
>
> If you say it makes no sense to document that stuff, I can't argue
> against that.  Sorry if you feel blamed (actually I didn't blame anyone)
> - but if this is to be "used at own risk" and not worth or possible to
> be documented decently, why has it been added to Emacs?

We can remove the option too then.  Personally, I am happy with the
graphical prompt, because it's more secure than entering passwords
through Emacs.

I added it to Emacs just because I got too many complaints from others
saying that Emacs integration of GnuPG password prompt is not user
friendly, as if it is a crucial part.  I don't think so, but for over a
decade I have made every possible attempt to improve the situation,
including pinentry.el, the loopback pinentry, pinentry-curses fallback,
though there is still no single reliable option.

If you want the feature with decent documentation, you should talk to
and work with the GnuPG upstream and your distro rather than whining
here.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Daiki Ueno <ueno@gnu.org>
> Cc: Eli Zaretskii <eliz@gnu.org>,  27445@debbugs.gnu.org
> Date: Fri, 24 Nov 2017 11:49:09 +0100
> 
> [...] whining here.

Please don't use such language here, certainly not when replying to
legitimate bug reports.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Daiki Ueno]

Eli Zaretskii <eliz@gnu.org> writes:

> Please don't use such language here, certainly not when replying to
> legitimate bug reports.

Legitimate bug reports should report the root problem to the relevant
project (GnuPG or Debian), instead of insisting on documenting a broken
behavior inherently caused by an external factor.

Otherwise, they even don't notice there is a problem.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Daiki Ueno <ueno@gnu.org>
> Cc: michael_heerdegen@web.de,  27445@debbugs.gnu.org
> Date: Fri, 24 Nov 2017 16:04:45 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Please don't use such language here, certainly not when replying to
> > legitimate bug reports.
> 
> Legitimate bug reports should report the root problem to the relevant
> project (GnuPG or Debian), instead of insisting on documenting a broken
> behavior inherently caused by an external factor.

Feel free to point that out, but without insulting the people who
report problems here.  We don't want to discourage users from
reporting issues because they are afraid of being publicly humiliated.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

  > We can remove the option too then.  Personally, I am happy with the
  > graphical prompt, because it's more secure than entering passwords
  > through Emacs.

Please keep Emacs and GOG working together on text terminals too.
That is how I use them.

-- 
Dr Richard Stallman
President, Free Software Foundation (https://gnu.org, https://fsf.org)
Internet Hall-of-Famer (https://internethalloffame.org)
Skype: No way! See https://stallman.org/skype.html.

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[diego]

I see pinentry.el was removed from emacs master - is it worth mentioning it did more than the functionality stated as making it obsolete?

For example, as long as INSIDE_EMACS was set, I could decrypt files using gpg from eshell or term and be prompted through the minibuffer, and git could sign commits while prompting through the minibuffer as well. 

I guess I'm saying it seems to me that pinentry.el had more use than just being used by epa. 

​- diego

​

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Rasmus]

Hi,

I also noticed that pinentry was removed, as I used to call pinentry-start
in my init.el.

Would it be worth mentioning the removal of pinentry in the NEWS file?  If
so, should the entry recommend simply set the epa-pinentry-mode instead,
probably to loopback?

Thanks,
Rasmus

bug#27445: 26.0.50; Insufficient documentation for pinentry.el

[Eli Zaretskii]

> From: Rasmus <rasmus@gmx.us>
> Date: Tue, 12 Dec 2017 10:19:02 +0100
> 
> I also noticed that pinentry was removed, as I used to call pinentry-start
> in my init.el.
> 
> Would it be worth mentioning the removal of pinentry in the NEWS file?  If
> so, should the entry recommend simply set the epa-pinentry-mode instead,
> probably to loopback?

Thanks, done.