Does service lookup by name work on Windows now?

Does service lookup by name work on Windows now?

[Robert Pluim]

The reason I ask is that gnus does this:

      (open-network-stream
       "*nnimap*" (current-buffer) nnimap-address
       (nnimap-map-port (car ports))

where

    (defun nnimap-map-port (port)
      (if (equal port "imaps")
          "993"
        port))

ostensibly because there were problems on Windows about 8 years ago
with using 'imaps' instead of the numeric value (I haven't managed to
track down the relevant discussion).

Iʼd like to remove this workaround, because it makes auth-source do
lookups using 993 rather than imaps when called from
open-network-stream, forcing users to use the numeric value in their
.authinfo.

Robert

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

How would one check if this works or doesn't work?

Re: Does service lookup by name work on Windows now?

[Robert Pluim]

Eli Zaretskii <eliz@gnu.org> writes:

> How would one check if this works or doesn't work?

If 

(open-network-stream "foo" nil "imap.gmail.com" "imaps")

works, it works. If it complains about 'service not found' or similar,
it doesnʼt.

Robert

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Sat, 26 Jan 2019 13:05:26 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > How would one check if this works or doesn't work?
> 
> If 
> 
> (open-network-stream "foo" nil "imap.gmail.com" "imaps")
> 
> works, it works. If it complains about 'service not found' or similar,
> it doesnʼt.

Works on Windows 8.1, doesn't work on XP.  Which is to be expected,
given the contents of etc/services on each system.

Re: Does service lookup by name work on Windows now?

[Robert Pluim]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Robert Pluim <rpluim@gmail.com>
>> Cc: emacs-devel@gnu.org
>> Date: Sat, 26 Jan 2019 13:05:26 +0100
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> > How would one check if this works or doesn't work?
>> 
>> If 
>> 
>> (open-network-stream "foo" nil "imap.gmail.com" "imaps")
>> 
>> works, it works. If it complains about 'service not found' or similar,
>> it doesnʼt.
>
> Works on Windows 8.1, doesn't work on XP.  Which is to be expected,
> given the contents of etc/services on each system.

XP has been end of life for 5 years. Iʼd really like to remove that
workaround, but I could leave it in, conditioned on (eq system-type
'windows-nt) or similar.

Robert

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Sat, 26 Jan 2019 13:34:50 +0100
> 
> > Works on Windows 8.1, doesn't work on XP.  Which is to be expected,
> > given the contents of etc/services on each system.
> 
> XP has been end of life for 5 years.

We don't follow orders from Microsoft.  We still support Windows 9X,
let alone XP.

> Iʼd really like to remove that workaround, but I could leave it in,
> conditioned on (eq system-type 'windows-nt) or similar.

Conditioning is fine with me.  We could also look at

  (car (x-server-version))

It should be 5 or less for XP and older systems.

Re: Does service lookup by name work on Windows now?

[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. ]]]

Each version of Windows was more repressive than the previous one.
If we are going to support Windows at all, we may as well support
the old versions that are still widely used.

-- 
Dr Richard Stallman
President, Free Software Foundation (https://gnu.org, https://fsf.org)
Internet Hall-of-Famer (https://internethalloffame.org)

Re: Does service lookup by name work on Windows now?

[Tim Cross]

[-- Attachment #1: Type: text/plain, Size: 1790 bytes --]

The only problem with continuing to support any OS which is no longer
maintained is that, in many cases, we are implicitly supporting users on
platforms with significant and well known exploits. This is especially true
with Windows XP, which has numerous, well documented and easy to exploit
vulnerabilities.

While I would agree that the decision to use a vulnerable OS is up to the
individual user, in practice, many users don't understand the risks and
consequences. To make it worse, exploited systems can also be a threat (or
inconvenience) to other users (for example, by being the source of botnets
or DDoS attacks).

There is an argument that it would be more responsible to not support any
OS once it is no longer maintained and receiving patches for security
vulnerabilities to discourage continued use of vulnerable systems and
encourage users to update to a current and more secure OS (which of course
could be GNU Linux!).  This may also make maintenance of Emacs easier as it
would reduce the need for exceptions and work arounds for older systems
which don';t support more modern OS practices.

Tim

On Sun, 27 Jan 2019 at 09:48, Richard Stallman <rms@gnu.org> wrote:

> [[[ 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. ]]]
>
> Each version of Windows was more repressive than the previous one.
> If we are going to support Windows at all, we may as well support
> the old versions that are still widely used.
>
> --
> Dr Richard Stallman
> President, Free Software Foundation (https://gnu.org, https://fsf.org)
> Internet Hall-of-Famer (https://internethalloffame.org)
>
>
>
>

-- 
regards,

Tim

--
Tim Cross

[-- Attachment #2: Type: text/html, Size: 2566 bytes --]

Re: Does service lookup by name work on Windows now?

[Alexandre Garreau]

[-- Attachment #1: Type: text/plain, Size: 3875 bytes --]

On 2019-01-27 at 16:11, Tim Cross wrote:
> The only problem with continuing to support any OS which is no longer
> maintained is that, in many cases, we are implicitly supporting users on
> platforms with significant and well known exploits. This is especially true
> with Windows XP, which has numerous, well documented and easy to exploit
> vulnerabilities.

If so it’d mean emacs would be supporting Windows at all, so it should
stop supporting it at all, which complies with what previously said.

However emacs is present on Windows to give a (imho, very good) taste of
what free software can be (in one of the best way of showing it), so to
advertise for other operating systems, so it’s not supporting anything
at all but freedom.

> While I would agree that the decision to use a vulnerable OS is up to the
> individual user, in practice, many users don't understand the risks and
> consequences. To make it worse, exploited systems can also be a threat (or
> inconvenience) to other users (for example, by being the source of botnets
> or DDoS attacks).

When a computer is under Windows XP it is often because it’s simply
impossible to switch.  It is then way better to allow most of (free)
software(s) to keep being updated on these, as otherwise it’s simply
that the old versions of these are going to be used instead, increasing
the attack surface.  So this is an argument supporting much more the
reverse: it would harm to make emacs stop supporting XP.

Also a lot of XP computers are ran on special hardware that allegedly
can’t be changed at all without more problem (I’ve heard “nuclear
plants” once, meanwhile I effectively observed XP being consistently
used in nuclear installations, though not active nuclear plants), and
that is consequently kept offline (yes there are offline attacks such as
worms, stuxnet, etc. but that’s also a question of policy).

Also while I regularely see Vista being used by many unaware uncaring
users, I didn’t see XP since a very long time ago except on computers of
more expecienced peoples that *had* the technical knowledge allowing
them to know there is indeed more crap in Vista than XP.  So I believe
the technical level of awareness is better for XP users than Vista users
in the end.

> There is an argument that it would be more responsible to not support any
> OS once it is no longer maintained and receiving patches for security
> vulnerabilities to discourage continued use of vulnerable systems and
> encourage users to update to a current and more secure OS (which of course
> could be GNU Linux!).  This may also make maintenance of Emacs easier as it
> would reduce the need for exceptions and work arounds for older systems
> which don';t support more modern OS practices.

Quirks are to be excepted from everywhere.  Especially from old
not-that-used UNIXes: Windows is much more a case of something
consistent with itself (since it’s completely opaque), and used by many
many many people (so it’s less work to make it compatible with more
environments).

It’s not only the fault of old systems but much more from diversity.
That is something you mostly find in more “open” (lots of commercial
UNIXes), free (all the free distributions), and imho is going to
increase in the future, albeit some other proprietary, highly insecure,
and widely as well as increasingly used non-GNU/Linux systems, such as
Android (which imho is going to be with time way worse than anything XP
could have been).

Btw it’s GNU/Linux, GNU–Linux, GNU-Linux, etc… maybe even “GNU” alone in
some cases …but not “GNU Linux”, since we are talking of the whole
operating system, not some “Linux” (which is a kernel), which would be
from GNU (and there is none, or you are confusing with GNU Linux-libre
maybe).

[-- Attachment #2.1: Type: text/html, Size: 4357 bytes --]

Re: Does service lookup by name work on Windows now?

[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. ]]]

I just wrote this response, which I will post in https://gnu.org/philosophy/.
Any comments or suggestions?


What Is the Right Way to Upgrade an Installation of Windows?


It is commonplace in the computing field to urge users to "upgrade" to
newer versions of Windows (and other nonfree programs) so as to get
fixes for "security".  This conclusion follows from the assumption
that these programs are honest software, designed to treat the user
right.  We do not expect that to be the case, and we know it is not
the case for Windows.  Therefore, we make a different recommendation.

In fact, newer versions of Windows have exploits before they are
released.  They are designed with various kinds of malicious
functionalities; see https://gnu.org/malware/malware-microsoft.html.

Newer Windows versions contain added malicious functionalities; they
give Microsoft more power over the users.  That is a reason not to
switch to a newer version.  For some users, switching to a newer
version may require rewarding Microsoft with money and even buying a
new computer.  We hardly want to encourage that!

Therefore we decline to support Microsoft by urging users to move to
newer versions; we do not treat the unintentional flaws of Windows as
more important than the intentional ones.

Our advice to those using any version of Windows is to upgrade to
GNU/Linux.

Supporting GNU packages on any version of Windows (or any non-GNU-like
systems) is not part of the GNU Project's core mission.  Our slogan
is, "It runs best on GNU/Linux."  We cooperate with the users who wish
to maintain that support, because we may as well cooperate when it is
not difficult.  We have no responsibility to continue doing so, but as
long as it is feasible and not holding us back, we have no reason to
stop.

-- 
Dr Richard Stallman
President, Free Software Foundation (https://gnu.org, https://fsf.org)
Internet Hall-of-Famer (https://internethalloffame.org)

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

> From: Richard Stallman <rms@gnu.org>
> Cc: eliz@gnu.org, rpluim@gmail.com, emacs-devel@gnu.org
> Date: Sun, 27 Jan 2019 18:34:10 -0500
> 
> I just wrote this response, which I will post in https://gnu.org/philosophy/.
> Any comments or suggestions?

Sounds fine to me.

Re: Does service lookup by name work on Windows now?

[Alexandre Garreau]

[-- Attachment #1: Type: text/plain, Size: 2289 bytes --]

So to make the thing not fully negative, would it be useful to recall
that sometimes we make free software work on Windows so that to offer a
taste of freedom, and state in this text that in when doing so, we
prefer to keep this software working on older versions?

Maybe something could be added on the fact that computers should be
owned by users and then it should be their choice to get new
(anti-)features or not (and to choose between stability and
flexibility)… be it, preferably, with “security updates” backported on
old versions, as it is only possible with free software, and as “some
GNU/Linux distributions” (so not to cite Debian) already do.

Then recalling it is nowadays commonplace for proprietary software to
force upgrades (often with multiple anti-features), that this is an
anti-feature, that this implies an universal backdoor, and that is a
good reason to go GNU/Linux instead of Windows 8/10, so you might hook
there the anti-windows-8/10 campaign.

Also something, maybe too much but still: forced upgrades are often
described by software editors as a “feature” because it allows to
correct “bugs” or “security holes” that are client-side: for instance, a
glitch in a game that allow for a user to cheat in their own interest,
or more commonly they notice spam, or harassement, begins to appear in
their “application”, so they “fortunately” and happily update so to
allow censorship *on the clients that cause the problem*.

So instead of having some clients capable to block other users, or to
detect cheat (which may be seen as a feature), they make the application
sending the spam unable to work or send anything (while this is still
technically possible if they knew the protocol), or the game unable to
cheat: in each case, the (anti-)“feature” is made so that to go against
the will of the user that gets the update.  This is also a wrong way of
approaching security, as it keeps protocols and receiver-clients (rather
than whole clients or their emitting part) dumb, insecure and inferior.
A thing that is impossible with free software (you just download the old
version, or change the software, and exploit the protocol again: so you
*have* to fix it reader-side, or protocol-side).

[-- Attachment #2.1: Type: text/html, Size: 2354 bytes --]

Re: Does service lookup by name work on Windows now?

[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. ]]]

  > So to make the thing not fully negative, would it be useful to recall
  > that sometimes we make free software work on Windows so that to offer a
  > taste of freedom, and state in this text that in when doing so, we
  > prefer to keep this software working on older versions?

It may be useful to mention that point.

  > Then recalling it is nowadays commonplace for proprietary software to
  > force upgrades (often with multiple anti-features), that this is an
  > anti-feature, that this implies an universal backdoor, and that is a
  > good reason to go GNU/Linux instead of Windows 8/10, so you might hook
  > there the anti-windows-8/10 campaign.

I think there is no need to spread it so far.

-- 
Dr Richard Stallman
President, Free Software Foundation (https://gnu.org, https://fsf.org)
Internet Hall-of-Famer (https://internethalloffame.org)

Re: Does service lookup by name work on Windows now?

[Robert Pluim]

[-- Attachment #1: Type: text/plain, Size: 353 bytes --]

Eli Zaretskii <eliz@gnu.org> writes:

>> Iʼd really like to remove that workaround, but I could leave it in,
>> conditioned on (eq system-type 'windows-nt) or similar.
>
> Conditioning is fine with me.  We could also look at
>
>   (car (x-server-version))
>
> It should be 5 or less for XP and older systems.

This is what I have so far.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Don-t-map-imaps-to-993-anymore-except-on-old-Windows.patch --]
[-- Type: text/x-patch, Size: 1071 bytes --]

From 696e57b862b9d525aae1a735f922025d8fa40bfd Mon Sep 17 00:00:00 2001
From: Robert Pluim <rpluim@gmail.com>
Date: Sun, 27 Jan 2019 16:13:46 +0100
Subject: [PATCH] Don't map imaps to 993 anymore except on old Windows versions
To: emacs-devel@gnu.org

'open-network-stream' will do the imaps service lookup itself, and
using 993 forced the user to use the numeric value in .authinfo for
certificate lookups.

* lisp/gnus/nnimap.el (nnimap-map-port): Only do mapping for Windows
XP or earlier.
---
 lisp/gnus/nnimap.el | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lisp/gnus/nnimap.el b/lisp/gnus/nnimap.el
index 75b5af8aab..9646bb51d0 100644
--- a/lisp/gnus/nnimap.el
+++ b/lisp/gnus/nnimap.el
@@ -413,8 +413,11 @@ nnimap-open-connection
 	nil
       stream)))
 
+;; This is only needed for Windows XP or earlier
 (defun nnimap-map-port (port)
-  (if (equal port "imaps")
+  (if (and (eq system-type 'windows-nt)
+           (<= (car (x-server-version)) 5)
+           (equal port "imaps"))
       "993"
     port))
 
-- 
2.20.1.142.g77556354bb

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Mon, 28 Jan 2019 11:12:15 +0100
> 
> This is what I have so far.

LGTM, but I hope someone from the Gnus team eyeballs this as well.

Thanks.

Re: Does service lookup by name work on Windows now?

[Robert Pluim]

Eli Zaretskii <eliz@gnu.org> writes:

>> This is what I have so far.
>
> LGTM, but I hope someone from the Gnus team eyeballs this as well.

Lars saw no problem with an earlier (unconditional) version of the
patch.

Robert

Re: Does service lookup by name work on Windows now?

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Gmane-Reply-To-List: yes
> Date: Mon, 04 Feb 2019 14:02:13 +0100
> Cc: Lars Ingebrigtsen <larsi@gnus.org>, emacs-devel@gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> This is what I have so far.
> >
> > LGTM, but I hope someone from the Gnus team eyeballs this as well.
> 
> Lars saw no problem with an earlier (unconditional) version of the
> patch.

Then I think we are good, thanks.

Re: Does service lookup by name work on Windows now?

[Robert Pluim]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Robert Pluim <rpluim@gmail.com>
>> Gmane-Reply-To-List: yes
>> Date: Mon, 04 Feb 2019 14:02:13 +0100
>> Cc: Lars Ingebrigtsen <larsi@gnus.org>, emacs-devel@gnu.org
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> >> This is what I have so far.
>> >
>> > LGTM, but I hope someone from the Gnus team eyeballs this as well.
>> 
>> Lars saw no problem with an earlier (unconditional) version of the
>> patch.
>
> Then I think we are good, thanks.

Thanks, pushed as f33a5dc947

Robert