gnutls error on Emacs for Android armeabi 7a

gnutls error on Emacs for Android armeabi 7a

[Lockywolf]

[-- Attachment #1: Type: text/plain, Size: 2431 bytes --]

Hello, everyone.

I am using Emacs on an Onyx Boox Max 2, which has Android 6.
Installed from SourceForge.

Most things work as expected, but I cannot install anything from Melpa.
And I also cannot use
(url-retrieve-synchronously "https://stable.melpa.org/packages/")

#+begin_src elisp
Debugger entered--Lisp error: (error "Invalid format character: ‘%t’")
  signal(error ("Invalid format character: ‘%t’"))
  error("Invalid format character: `%%%c'" 116)
  format-spec("gnutls-cli --x509cafile %t -p %p %h" ((116) (104 . "stable.melpa.org") (112 . "443")))
  open-tls-stream("stable.melpa.org" #<buffer  *url-http-temp*> "stable.melpa.org" 443)
  network-stream-open-tls("stable.melpa.org" #<buffer  *url-http-temp*> "stable.melpa.org" 443 (:type tls :nowait nil))
  open-network-stream("stable.melpa.org" #<buffer  *url-http-temp*> "stable.melpa.org" 443 :type tls :nowait nil)
  url-open-stream("stable.melpa.org" #<buffer  *url-http-temp*> "stable.melpa.org" 443 tls)
  url-http-find-free-connection("stable.melpa.org" 443 tls)
  url-http(#s(url :type "https" :user nil :password nil :host "stable.melpa.org" :portspec nil :filename "/packages/" :target nil :attributes nil :fullness t :silent nil :use-cookies t :asynchronous nil) #f(compiled-function (&rest args) #<bytecode -0x1c8f7ded>) (nil) nil tls)
  url-https(#s(url :type "https" :user nil :password nil :host "stable.melpa.org" :portspec nil :filename "/packages/" :target nil :attributes nil :fullness t :silent nil :use-cookies t :asynchronous nil) #f(compiled-function (&rest args) #<bytecode -0x1c8f7ded>) (nil))
  url-retrieve-internal("https://stable.melpa.org/packages/" #f(compiled-function (&rest args) #<bytecode -0x1c8f7ded>) (nil) nil nil)
  url-retrieve("https://stable.melpa.org/packages/" #f(compiled-function (&rest args) #<bytecode -0x1c8f7ded>) nil nil nil)
  url-retrieve-synchronously("https://stable.melpa.org/packages/")
  (progn (url-retrieve-synchronously "https://stable.melpa.org/packages/"))
  eval((progn (url-retrieve-synchronously "https://stable.melpa.org/packages/")) t)
  elisp--eval-last-sexp(nil)
  eval-last-sexp(nil)
  funcall-interactively(eval-last-sexp nil)
  call-interactively(eval-last-sexp nil nil)
  command-execute(eval-last-sexp)
#+end_src

Has anyone seen anything like that?
Any ideas how to fix this?

-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: gnutls error on Emacs for Android armeabi 7a

[Po Lu]

Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
writes:

> Hello, everyone.
>
> I am using Emacs on an Onyx Boox Max 2, which has Android 6.
> Installed from SourceForge.

Which package did you download?

> Most things work as expected, but I cannot install anything from Melpa.
> And I also cannot use
> (url-retrieve-synchronously "https://stable.melpa.org/packages/")

This suggests that Emacs was configured without TLS support.  These
packages support GnuTLS:

  emacs-30.0.50-21-arm64-v8a.apk
  emacs-30.0.50-29-arm64-v8a.apk
  emacs-30.0.50-21-x86_64.apk
  emacs-30.0.50-29-x86_64.apk
  emacs-30.0.50-16-armeabi-v7a.apk
  emacs-30.0.50-24-armeabi-v7a.apk

the remainder do not, though I plan to gradually port GnuTLS to the
systems concerned in due course.

Re: gnutls error on Emacs for Android armeabi 7a

[Lockywolf]

[-- Attachment #1: Type: text/plain, Size: 1302 bytes --]


Po Lu <luangruo@yahoo.com> writes:

> Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
> writes:
>
>> Hello, everyone.
>>
>> I am using Emacs on an Onyx Boox Max 2, which has Android 6.
>> Installed from SourceForge.
>
> Which package did you download?
>

emacs-30.0.50-16-armeabi-v7a.apk

>> Most things work as expected, but I cannot install anything from Melpa.
>> And I also cannot use
>> (url-retrieve-synchronously "https://stable.melpa.org/packages/")
>
> This suggests that Emacs was configured without TLS support.  These
> packages support GnuTLS:
>
>   emacs-30.0.50-21-arm64-v8a.apk
>   emacs-30.0.50-29-arm64-v8a.apk
>   emacs-30.0.50-21-x86_64.apk
>   emacs-30.0.50-29-x86_64.apk
>   emacs-30.0.50-16-armeabi-v7a.apk
>   emacs-30.0.50-24-armeabi-v7a.apk
>
> the remainder do not, though I plan to gradually port GnuTLS to the
> systems concerned in due course.

This was emacs-30.0.50-16-armeabi-v7a.apk

>As regards libxml2, what gives you the impression that it's omitted
>from this package?

The fact that running (libxml2-available-p) returns nil.
And that opening epub files with nov.el results in it complaining about
libxml2 functions being void.


-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: gnutls error on Emacs for Android armeabi 7a

[Lockywolf]

[-- Attachment #1: Type: text/plain, Size: 717 bytes --]


Disregard all of the above, I have re-installed the most recent Emacs,
and now everything works, both gnutls and libxml2.

On the other hand, may I ask the builds for termux-16 be restored?

The "new" termux 0.119 has reinstated support for Android 5 and 6, and I
would greatly enjoy having access to termux from Emacs. (I guess, I can
re-pack it manually though.)

Or, maybe there is some other way of supporting zip/unzip on stupid
Android systems, which only have a busybox of unzip, but no zip at all.

Po Lu <luangruo@yahoo.com> writes:

> Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
> writes:
>


-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: gnutls error on Emacs for Android armeabi 7a

[Po Lu]

Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
writes:

> Disregard all of the above, I have re-installed the most recent Emacs,
> and now everything works, both gnutls and libxml2.
>
> On the other hand, may I ask the builds for termux-16 be restored?
>
> The "new" termux 0.119 has reinstated support for Android 5 and 6, and I
> would greatly enjoy having access to termux from Emacs. (I guess, I can
> re-pack it manually though.)

I think you're a victim of misinformation, since the Termux repository
states in no uncertain terms:

  Latest version is v0.118.0.

  NOTICE: It is highly recommended that you update to v0.118.0 or higher
  ASAP for various bug fixes, including a critical world-readable
  vulnerability reported here.  Also reminding again to users who have
  installed termux apps from google playstore that playstore builds are
  deprecated and no longer supported.  It is recommended that you shift to
  F-Droid or GitHub releases.

  Termux can be obtained through various sources listed below for only
  Android >= 7 with full support for apps and packages.

  Support for both app and packages was dropped for Android 5 and 6 on
  2020-01-01 at v0.83, however it was re-added just for the app without
  any support for package updates on 2022-05-24 via the GitHub sources.
  Check here for the details.

and a search revealed also this issue/discussion:

  https://github.com/termux/termux-app/discussions/3812

in which there are several strong words of caution from its maintainers
against packages purporting to be 0.119.  Since packaging support for
Android 5 and 6 remains inactive, and the limited support they do
provide was not functional when last I tried, I don't plan to continue
providing builds for such configurations until it is demonstrated that
they are in fact practical.  Recall that the recently superseded Termux
builds labeled "16" enforced a minimum API level of 24 in any case,
which state of affairs continued for a considerable period without any
reports from users of having been inconvenienced.

> Or, maybe there is some other way of supporting zip/unzip on stupid
> Android systems, which only have a busybox of unzip, but no zip at all.

If it's only nov.el that you need,

  (setq nov-unzip-args '("-o" filename "-d" directory))

the built-in arc-mode should detect busybox or Android unzip and adjust
its command line unassisted, but if not, please report the failure as a
bug.

Re: gnutls error on Emacs for Android armeabi 7a

[Lockywolf]

[-- Attachment #1: Type: text/plain, Size: 1863 bytes --]


Po Lu <luangruo@yahoo.com> writes:

1.
> and a search revealed also this issue/discussion:
>
>   https://github.com/termux/termux-app/discussions/3812
>
> in which there are several strong words of caution from its maintainers
> against packages purporting to be 0.119.  Since packaging support for
> Android 5 and 6 remains inactive, and the limited support they do
> provide was not functional when last I tried, I don't plan to continue
> providing builds for such configurations until it is demonstrated that
> they are in fact practical.  Recall that the recently superseded Termux
> builds labeled "16" enforced a minimum API level of 24 in any case,
> which state of affairs continued for a considerable period without any
> reports from users of having been inconvenienced.

I installed termux from GitHub workflow, not from some UltraMod website.
It works for my purposes, more or less.
I have also created a self-signed pair of packages for termux, and
emacs. I didn't actually rebuild them, I have just unpacked and
re-signed them with apktool/jarsigner.
(Like here https://github.com/marek-g/marek-g.github.io/blob/44d1b74493e35ef9f2e206657f42d3e927ee7751/posts/tips_and_tricks/emacs_on_android.md)

Well, unmaintained, unsupported, buggy... but still lets me use my
7-years-old ebook reader almost as a true laptop, with git and gcc, and
now with Emacs.

2. Unrelated to termux, but related to Android.

Is there a way to stop the on-screen keyboard pop-up when C-s or C-r is
pressed? I am using a physical keyboard, so I don't really need the
on-screen one.

It does _not_ pop up when I use any other keyboard-involving command,
such as M-x, or even non-incremental search (spawned when I type the 🔍
button present on some keyboards).

-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: gnutls error on Emacs for Android armeabi 7a

[Po Lu]

Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
writes:

> I installed termux from GitHub workflow, not from some UltraMod website.
> It works for my purposes, more or less.
> I have also created a self-signed pair of packages for termux, and
> emacs. I didn't actually rebuild them, I have just unpacked and
> re-signed them with apktool/jarsigner.
> (Like here
> https://github.com/marek-g/marek-g.github.io/blob/44d1b74493e35ef9f2e206657f42d3e927ee7751/posts/tips_and_tricks/emacs_on_android.md)
>
> Well, unmaintained, unsupported, buggy... but still lets me use my
> 7-years-old ebook reader almost as a true laptop, with git and gcc, and
> now with Emacs.

The Termux binaries on SourceForge are signed with public signing keys
published in the Emacs repository and meant for the resigned Termux
being distributed from the same location.  Users signing builds
individually should either compile Emacs from source code, with a
different set of signing keys, or modify any Emacs package they please,
not limited to what is in the "termux" directory.

Generally speaking, it is redundant to generate a signing key and resign
both Termux and Emacs, as their default keys are available to the
public:

  https://git.savannah.gnu.org/cgit/emacs.git/tree/java/emacs.keystore
  https://github.com/termux/termux-app/blob/master/app/testkey_untrusted.jks

since, evidently, Android's policy of restricting program
interoperability to programs from a single publisher does not commend
itself to us or them.

> 2. Unrelated to termux, but related to Android.
>
> Is there a way to stop the on-screen keyboard pop-up when C-s or C-r is
> pressed? I am using a physical keyboard, so I don't really need the
> on-screen one.

Disable "Settings -> Language & Input -> Physical keyboard -> Use
on-screen keyboard."

Re: gnutls error on Emacs for Android armeabi 7a

[Lockywolf]

[-- Attachment #1: Type: text/plain, Size: 1052 bytes --]


Po Lu <luangruo@yahoo.com> writes:

>> 2. Unrelated to termux, but related to Android.
>>
>> Is there a way to stop the on-screen keyboard pop-up when C-s or C-r is
>> pressed? I am using a physical keyboard, so I don't really need the
>> on-screen one.
>
> Disable "Settings -> Language & Input -> Physical keyboard -> Use
> on-screen keyboard."

No such option on my ebook reader.

I can configure the physical keyboard using the persistent notification,
which lists input method apps, and allows changing layouts for the
physical keyboard, but there is no "Use on-screen keyboard" option to
un-check.

Moreover, the on-screen keyboard does not appear when pressing M-x, or
M-:, so this seems to be somehow search-related. Moreover, I can press
"back" to turn the on-screen keyboard off, which is, like, just one more
keypress. It's just annoying.

And I wouldn't want to disable the on-screen keyboard entirely, after
all, I am using it "sometimes".

-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: gnutls error on Emacs for Android armeabi 7a

[Po Lu]

Lockywolf <for_help-gnu-emacs_at_gnu_org_2023-11-30@lockywolf.net>
writes:

> No such option on my ebook reader.
>
> I can configure the physical keyboard using the persistent notification,
> which lists input method apps, and allows changing layouts for the
> physical keyboard, but there is no "Use on-screen keyboard" option to
> un-check.

If a keyboard icon is displayed on the navigation bar during Isearch,
the popup it activates should provide a toggle titled "Hardware
keyboard--show input method" or thereabouts, which is equivalent to this
Settings option.

> Moreover, the on-screen keyboard does not appear when pressing M-x, or
> M-:, so this seems to be somehow search-related. Moreover, I can press
> "back" to turn the on-screen keyboard off, which is, like, just one more
> keypress. It's just annoying.

The issue is that Emacs knows nothing of whether the on-screen keyboard
is displayed, or indeed, whether such a keyboard is available at all.
It notifies the system after a command that sends Emacs into a state
where keyboard input is required, and leaves the decision whether to
display the keyboard entirely in its hands.