Jay Belanger | 19 Jan 00:01 2016
Picon

Odd spacing issues


I have SXEmacs installed on my laptop and my desktop computers.
If I run Calc on my desktop (installed from ftp.pasteur.fr in
pub/computing/xemacs/beta/experimental/packages), I get odd
spacing; for example:
  --- Emacs Calculator Mode ---
  3:  1

  2:  2

  1:  3

If I run it from my laptop, the spacing isn't there:
  --- Emacs Calculator Mode ---
  3:  1
  2:  2
  1:  3

Both computers are running the latest git version of SXEmacs, and their
packages and sxemacs directories are kept in sync.  They might
have been built with some different libraries.

I may look into why there is a difference when I get a chance, but
meanwhile does anybody have any ideas?

On an unrelated note, M-x term is very buggy for me.
(For example, when I start it, I get an error:
  (1) (error/warning) Error in process filter: (error /home/belange/ is not a directory)
My home directory is /home/belanger/, so an "r" is missing.  There are also
problems with horizontal spacing.)  This happens for both computers.
(Continue reading)

issue-tracking | 12 Jan 03:42 2016

[Bug 180] New: No audio output from certain WAV formats using FFmpeg media backend

Bug ID Summary Product Version Hardware OS Status Severity Priority Component Assignee Reporter QA Contact
180
No audio output from certain WAV formats using FFmpeg media backend
SXEmacs
22.1.15
PC
Linux
NEW
normal
P5
General
njsf <at> sxemacs.org
steve <at> sxemacs.org
sxemacs-devel <at> sxemacs.org

Created attachment 173 [details] Test files The FFmpeg media backend will play WAV files, but WAVs of certain formats do not output any (or almost none) audio. For the purpose of testing I took 'login.wav' from Pidgin (/usr/share/sounds/purple/login.wav), and using SoX created a set of WAVs with different channels/rate/depth. The test files are attached. The test files all play as expected with 'ffplay', 'play' (SoX), 'aplay' (ALSA), 'paplay' (PulseAudio), and 'mplayer'. Recipe: 1) Unpack the testfiles.tar.xz somewhere 2) In a -no-autoloads instance, eval the following in *scratch*... (progn (load "sound") (let ((wavs (directory-files "/your/path/to/testfiles" t "wav$" nil t)) (dev (make-audio-device 'pulse))) (mapcar #'(lambda (wav) (let ((stream (make-media-stream :file wav 'ffmpeg))) (message "Trying: %S" wav) (play-media-stream-synchronously stream dev))) wavs))) You don't have to use pulse, other device types (alsa, ao, etc) will give the same results. For me, the WAVs at 8000 Hz didn't play properly, producing no, or very little audio output. The stereo 8000 Hz WAVs would output a tiny portion of the audio right at the end of the stream, while the mono ones output nothing other than static. All the test files will play properly if you use a different media backend, such as 'sox or 'sndfile.
You are receiving this mail because:
  • You are the QA Contact for the bug.
issue-tracking | 5 Jan 13:21 2016

[Bug 179] New: Cannot play ogg or mp3 streams using FFmpeg media backend

Bug ID Summary Product Version Hardware OS Status Severity Priority Component Assignee Reporter QA Contact
179
Cannot play ogg or mp3 streams using FFmpeg media backend
SXEmacs
22.1.15
PC
Linux
NEW
normal
P5
General
njsf <at> sxemacs.org
steve <at> sxemacs.org
sxemacs-devel <at> sxemacs.org

Created attachment 171 [details] backtrace I have tried several ogg and mp3 files, all of which play fine with all other audio tools, including ffplay. But SXEmacs will not play any ogg or mp3. In a -no-autoloads instance... (load "sound") (let ((stream (make-media-stream :file "/path/to/file.ogg" 'ffmpeg)) (device (make-audio-device 'pulse))) (play-media-stream stream device)) Results in a SIGFPE arithmetic exception in media_ffmpeg_read(). Backtrace attached. Same happens for mp3
You are receiving this mail because:
  • You are the QA Contact for the bug.
Steve Youngs | 5 Jan 07:45 2016
X-Face
Face
Gravatar

Our (upcoming) Packages Repo

Hey Everyone!

OK, so I'm going to set up a packages repo, but before I go ahead with
that I wanted to discuss the best way to go about it.

My initial thought is to have a main git repo containing what's in the
toplevel of the XEmacs pkgs mercurial repo (basically, the build
infrastructure and aux files etc).  It would also include the
'xemacs-packages' and 'mule-packages' directories, but nothing beneath
that. Then each individual package would be a "sub-repo" under
{xemacs,mule}-packages. 

Does that sound feasible?  Please note that I don't know anything about
git sub-repos yet, I've never used them.

With the way the XEmacs packages source build infrastructure is currently
set up it doesn't make sense to checkout anything other than the entire
tree.  But I would like to allow for individual package checkouts in the
future.  Ideally, if you were to grab just one package it'd be nice if all
the dependency packages came along with it.  No idea how you'd go about
setting that up, or if it were even possible.

How do we make it modular and not end up in dependency hell?

I would also like to have it possible for package maintainers to be able
to write to the repo (push their own updates)[1] because I sure don't
want to maintain or be the gatekeeper to 135 packages.

So, what do you think?  Anyone got any alternate ideas/suggestions?  I'm
thinking about doing this on the weekend unless somebody has a better
plan. 

Footnotes: 
[1]  At this point, I don't know how I'd go about setting that up in a
secure fashion without giving pkg maintainers access to all of my stuff
on my host.

--

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|       SXEmacs - The only _______ you'll ever need.       |
|         Fill in the blank, yes, it's THAT good!          |
|------------------------------------<steve <at> sxemacs.org>---|
Steve Youngs | 5 Jan 01:47 2016
X-Face
Face
Gravatar

Re: [gnus] branch SYgnus updated (b83f807 -> 0dbf44d)

Steve Youngs <steve <at> sxemacs.org> writes:

  > Steve pushed a change to branch SYgnus
  > in repository gnus.

  >       from  b83f807   Merge branch 'master' of http://git.gnus.org/gnus into SYgnus
  >        new  b8d320e   Refactor mml-smime.el, mml1991.el, mml2015.el
  >        new  7067d9b   Apply patch by Katsumi Yamaoka  <yamaoka <at> jpl.org>
  >        new  f017906   gnustest-mml-sec.el: Remove require cl.
  >        new  16cb769   Remove lisp/tests/mml-gpghome/random_seed from version control
  >        new  437c079   * mml-sec.el (mml-secure-cust-record-keys):   Make gnus-union use `equal' to compare items
in lists.
  >        new  0dbf44d   Merge branch 'master' of http://git.gnus.org/gnus into SYgnus

I haven't tried building this yet.  If someone wants to grab it, build
it, and let me know of any issues...

--

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|       SXEmacs - The only _______ you'll ever need.       |
|         Fill in the blank, yes, it's THAT good!          |
|------------------------------------<steve <at> sxemacs.org>---|
Jay Belanger | 3 Jan 21:26 2016
Picon

Calc


Hi Everybody,

I took the version of Calc from GNU/Emacs and largely got
it working with SXEmacs.  I also started with the Calc manual
from XEmacs and added to it to reflect the changes, but 
I still need to add some more to the manual and edit it.

This version is at 
http://vh213601.truman.edu/~belanger/calc.tar

Next week I'll add more to the manual, and then
go through it to make sure everything is working.
Then I'll make a package out of it, in case anyone
is interested.  After that, I intend to change it 
to use extended numbers.

calc-keyboard (which turns a buffer into a xcalc-like
point-and-click calculator) doesn't work, and I didn't
bother to look at it.  But then, in the current XEmacs 
Calc, calc-keyboard isn't working right (at least not 
for me).  Unless someone tells me they're interested
in this version of Calc and also want calc-keyboard,
I'll probably just remove it.

Jay

issue-tracking | 2 Jan 04:24 2016

[Bug 178] New: Cannot load ffi-wand after creating an audio device in a FFmpeg-enabled SXEmacs

Bug ID Summary Product Version Hardware OS Status Severity Priority Component Assignee Reporter QA Contact
178
Cannot load ffi-wand after creating an audio device in a FFmpeg-enabled SXEmacs
SXEmacs
22.1.15
PC
Linux
NEW
normal
P5
General
njsf <at> sxemacs.org
steve <at> sxemacs.org
sxemacs-devel <at> sxemacs.org

Prerequisites: Current SXEmacs git (0e87679e6dc85cdadb5bd3dd0e6e273a2646c150) built with ffmpeg (--with-media=ffmpeg (or all)), and FFI, of course. And I think OpenSSL too. ImageMagick v6.6.9 OpenSSL v1.0.2e FFmpeg version info... libavutil 54. 27.100 / 54. 27.100 libavcodec 56. 41.100 / 56. 41.100 libavformat 56. 36.100 / 56. 36.100 libavdevice 56. 4.100 / 56. 4.100 libavfilter 5. 17.100 / 5. 17.100 libswscale 3. 1.101 / 3. 1.101 libswresample 1. 2.100 / 1. 2.100 libpostproc 53. 3.100 / 53. 3.100 Recipe: In a -no-autoloads session, eval in *scratch* (progn (load "sound") (require 'ffi) (make-audio-device 'pulse) ; same results with 'alsa', and 'ao' (ffi-load "libMagickWand")) The error returned is: Can't load library `libMagickWand': dlopen: cannot load any more object with static TLS There are currently two ways to avoid this problem... 1) Build SXEmacs without ffmpeg (--with-media=all,noffmpeg) 2) Ensure that you load ffi-wand _BEFORE_ creating any audio devices. For example, in that snippet above, if you swap the order of #'make-audio-device and #'ffi-load everything will load fine and there'll be no errors. ---------------------------------------------- I build everything from source (LFS) so here are details of my ImageMagick, OpenSSL, and FFmpeg (configure options, etc) ImageMagick: ./configure --prefix=/usr \ --infodir=/usr/share/info \ --mandir=/usr/share/man \ --sysconfdir=/etc \ --libexecdir=/usr/lib \ --localstatedir=/var \ --with-perl \ --disable-nls OpenSSl: ./config --prefix=/usr \ --openssldir=/etc/ssl \ shared \ zlib-dynamic FFmpeg: CPPFLAGS=-I/usr/X11R6/include \ LDFLAGS=-L/usr/X11R6/lib \ ${SRCTREE}/configure --prefix=/usr \ --enable-shared \ --enable-gpl \ --enable-nonfree \ --enable-postproc \ --enable-ffmpeg \ --enable-ffplay \ --enable-ffprobe \ --enable-ffserver \ --enable-avdevice \ --enable-avcodec \ --enable-avformat \ --enable-swscale \ --enable-avfilter \ --enable-pthreads \ --enable-network \ --enable-x11grab \ --enable-libfaac \ --enable-libfreetype \ --enable-libmp3lame \ --enable-libopenjpeg \ --enable-libpulse \ --enable-libspeex \ --enable-libtheora \ --enable-libvorbis \ --enable-libvpx \ --enable-libx264 \ --enable-libx265 \ --enable-libxvid \ --enable-doc
You are receiving this mail because:
  • You are the QA Contact for the bug.
Steve Youngs | 1 Jan 12:15 2016
X-Face
Face
Gravatar

FFmpeg issues

Hi Nelson!

Here is a run-down of the issues I'm seeing with FFmpeg.

First of all, a couple things I'd file under 'Non-critical, but we
should deal with it "one day"'

avcodec_decode_audio4() needs to be used instead of audio3, (anything
below 4 is deprecated).  I don't think audio2 even exists anymore so no
point in checking for it in configure, or having it in the code.

I'm sure you are completely aware of that. :-)

---

And now for something weird and baffling...

(progn
  (require 'ffi-wand)
  (make-audio-device 'pulse))

Works fine, and as expected.

(progn
  (make-audio-device 'pulse)
  (require 'ffi-wand))

Results in an error

Can't load library `libGraphicsMagickWand': libGraphicsMagickWand.so: cannot open shared object
file: No such file or directory

I get that because ffi-wand tries 3 different libs (to cover different
ImageMagick versions) in succession, failing gracefully on all but the
last.  My wand lib is tried before libGraphicsMagickWand, so this isn't
the real error.

(progn
  (make-audio-device 'pulse)
  (require 'ffi)
  (ffi-load "libMagickWand"))

This is...

Can't load library `libMagickWand': dlopen: cannot load any more object with static TLS

I also tried with (make-audio-device 'alsa), and
(make-audio-device 'ao)... same result.

---

And finally, using it...

I took TuxedoMoon.Jingle4.{au,wav} from XEmacs package and created a
.ogg, .mp3, and .flac with /usr/bin/ffmpeg.  And then started testing in
a -no-autoloads instance from gdb.

Here is my quick'n'dirty test harness

(load "sound")

(setq file-alist '((au . "file.au")
		   (wav . "file.wav")
		   (mp3 . "file.mp3")
		   (ogg . "file.ogg")
		   (flac . "file.flac")))

(defun sy-ff-test (type)
  (interactive "SFile type: ")
  (let* ((dir (paths-construct-path 
	       (list (user-home-directory) "tmp" "testfiles")))
	 (dev (make-audio-device 'pulse
				 :client "FF::Test"
				 :stream "FF::Stream"
				 :role "video"))
	 (file (expand-file-name (cdr (assq type file-alist)) dir))
	 (stream (make-media-stream :file file)))
    (play-media-stream stream dev)))
;; video role because that will cork my music

The results:

        au      Plays fine
        wav     No error, but no audio (plays fine with other tools).
                The odd thing here is that I was certain it played other
                wav's without issue.
        flac    Plays fine
        mp3     Program received signal SIGFPE, Arithmetic exception.
        ogg     Program received signal SIGFPE, Arithmetic exception.
                Both the ogg, and mp3 play fine with other tools

Same bt for both mp3 and ogg...

(bastard-gdb) bt
#0  0x00000000006a2de4 in media_ffmpeg_read (mss=<optimized out>, outbuf=0x29ee6d0,
length=<optimized out>) at /usr/src/sxemacs/sxemacs/src/media/media-ffmpeg.c:729
#1  0x00000000006a5a81 in stream_write_callback (stream=0x29ed150, length=14400,
userdata=0x2946970) at /usr/src/sxemacs/sxemacs/src/media/sound-pulse.c:717
#2  0x00007ffff53946b7 in create_stream_complete (s=0x29ed150) at /usr/src/pulseaudio/pulseaudio/src/pulse/stream.c:969
#3  pa_create_stream_callback (pd=pd <at> entry=0x7fffe0001590, command=command <at> entry=2,
tag=tag <at> entry=3, t=t <at> entry=0x7fffe0002690, userdata=userdata <at> entry=0x29ed150) at /usr/src/pulseaudio/pulseaudio/src/pulse/stream.c:1167
#4  0x00007ffff5124641 in run_action (pd=0x7fffe0001590, r=0x7fffe0006c60, command=2,
ts=0x7fffe0002690) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/pdispatch.c:284
#5  0x00007ffff51249a3 in pa_pdispatch_run (pd=0x7fffe0001590,
packet=packet <at> entry=0x7fffe0003970, ancil_data=ancil_data <at> entry=0x7fffe0000bf0,
userdata=userdata <at> entry=0x29187d0) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/pdispatch.c:337
#6  0x00007ffff53764be in pstream_packet_callback (p=<optimized out>, packet=0x7fffe0003970,
ancil_data=0x7fffe0000bf0, userdata=0x29187d0) at /usr/src/pulseaudio/pulseaudio/src/pulse/context.c:338
#7  0x00007ffff512709f in do_read (p=p <at> entry=0x7fffe0000960, re=re <at> entry=0x7fffe0000b28) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/pstream.c:870
#8  0x00007ffff51296db in do_pstream_read_write (p=0x7fffe0000960) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/pstream.c:189
#9  0x00007ffff5129944 in srb_callback (srb=<optimized out>, userdata=0x7fffe0000960) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/pstream.c:225
#10 0x00007ffff512a27a in srbchannel_rwloop (sr=0x7fffe00093a0) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/srbchannel.c:190
#11 0x00007ffff538a48f in dispatch_pollfds (m=0x2918620) at /usr/src/pulseaudio/pulseaudio/src/pulse/mainloop.c:655
#12 pa_mainloop_dispatch (m=m <at> entry=0x2918620) at /usr/src/pulseaudio/pulseaudio/src/pulse/mainloop.c:898
#13 0x00007ffff538a87c in pa_mainloop_iterate (m=0x2918620, block=<optimized out>, retval=0x0) at /usr/src/pulseaudio/pulseaudio/src/pulse/mainloop.c:929
#14 0x00007ffff538a920 in pa_mainloop_run (m=0x2918620, retval=retval <at> entry=0x0) at /usr/src/pulseaudio/pulseaudio/src/pulse/mainloop.c:944
#15 0x00007ffff5398603 in thread (userdata=0x29185e0) at /usr/src/pulseaudio/pulseaudio/src/pulse/thread-mainloop.c:88
#16 0x00007ffff5139a58 in internal_thread_func (userdata=0x291abb0) at /usr/src/pulseaudio/pulseaudio/src/pulsecore/thread-posix.c:81
#17 0x00007fffece59084 in start_thread (arg=0x7fffe4dbc700) at pthread_create.c:309
#18 0x00007fffed35cead in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111
(bastard-gdb) 

OK, I hope that is enough for you to go on.  Let me know if there is
anything more you need.

--

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|       SXEmacs - The only _______ you'll ever need.       |
|         Fill in the blank, yes, it's THAT good!          |
|------------------------------------<steve <at> sxemacs.org>---|
Steve Youngs | 31 Dec 18:40 2015
X-Face
Face
Gravatar

Fwd: Moving Gnus development to Emacs?

One that didn't find its way to us...

Will have to keep an eye on his repo.

Face Picon Gravatar
From: Lars Magne Ingebrigtsen <larsi <at> gnus.org>
Subject: Re: Moving Gnus development to Emacs?
Newsgroups: gmane.emacs.devel
Date: 2015-12-31 17:13:52 GMT
Uwe Brauer <oub <at> mat.ucm.es> writes:

> Although I switched mostly to GNU emacs, that decision is a bit sad.
> Could you at least tag the last Xemacs compat version, before removing
> that code?

I'll do one final release before closing git.gnus.org.  If that's what I
decide to do.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no


--

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|       SXEmacs - The only _______ you'll ever need.       |
|         Fill in the blank, yes, it's THAT good!          |
|------------------------------------<steve <at> sxemacs.org>---|
Steve Youngs | 31 Dec 18:15 2015
X-Face
Face
Gravatar

Looks Like We're On Our Own

Hey Folks!

I'm writing this a few minutes before 2016 so I hope you're all spending
quality time with those you love most to see out the last moments of
2015.  I'm not the partying type, and if it wasn't for the fact that my
wife hasn't yet returned home from work I would have been in bed hours
ago. :-)

On to something a little sombre.  In the last 24 hours I have been made
aware of three (3) fairly major elisp packages that are planning to, not
only stop supporting XEmacs, but to remove the compatibility code that
they already have.  The packages in question: Gnus, TRAMP, and CEDET.

As you saw, Lars (Mr Gnus) posted yesterday that he wanted to move his
source repo into Emacs'.  Yes, that will be a pain for anyone wanting to
track Gnus development, but not insurmountable.  The real kick in the
pants was in the same message he said he'd be removing all compat code.

I've been watching the thread on emacs-devel list because even though
Lars Cc'd us, not everyone responding kept the Cc header in place.  The
result?  The vast majority agree with, and support, his plans.  In
fact, only 2 people questioned it or disagreed (Jay and I, if you were
wondering).

How does this new trend of proactively hampering (even crushing) any and
all emacsen that aren't the exalted GNU/Emacs affect us?  Well, believe
it or not, but in the immediate and short terms it won't really affect
us at all.  The elisp packages we have today will still work tomorrow.
We won't see any impact until people take it upon themselves to update
existing packages to upstream versions.

Moving into the future, I think we should...

Immediate / short-term (30 days):

  o Set up a packages repo populated with the current content of the
    XEmacs package repo.

  o Begin the process of having SXEmacs-built binary packages available
    for d/l and install.

Medium-term (6 - 12 months):

  o Work on updating the dinosaur era code in the packages bringing them
    into the 21st century.  Or as close as we can to current upstream
    without running into the above lets-nuke-all-compat-code nonsense.

    I do understand how much of a mammoth task that is.  I'm not
    suggesting that this is a task that gets finished in the next 12
    months, but I would like to see decent progress.  I'd love to be
    releasing package updates every few days. :-)

  o Begin porting some of the more commonly used things from GNU/Emacs.
    And I think the first thing should probably be the CL stuff (that in
    itself would go a long way toward bridging the compatibility gap).

  o Begin porting some of the XEmacs 21.5 things we find to be
    appealing.  I've got nothing concrete in mind here, but there's
    bound to be some goodies there we'd like to have.

  o Depending on how the XEmacs team is travelling at this point, maybe
    offer to take over package distribution and source management.

Long-term (2 - 5 years):

  o Achieve full elisp compatibility with GNU/Emacs.  What I hope for
    is that if a .el runs out-of-the-box on GNU/Emacs, it should run
    out-of-the-box on SXEmacs.

    Right from the very beginning of this project I have said that
    SXEmacs will always be 100% compatible with XEmacs where elisp is
    concerned. Because of things that have unfolded recently (XEmacs
    pretty much giving up; GNU/attitude toward compat code) I now
    believe a wiser choice is to shoot for GNU/Emacs compatibility for
    elisp. 

  o All packages up to date with upstream stable versions, or better.

Don't worry, I'm not selling out, I have no desire or intention of
turning SXEmacs into a GNU/Emacs clone.  SXEmacs will always be
SXEmacs. :-)

Oh, and those above short/medium/long term goals, they don't replace any
of our pre-existing plans, they are in addition to and in parallel with
our existing plans.

Happy New Year!

--

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|       SXEmacs - The only _______ you'll ever need.       |
|         Fill in the blank, yes, it's THAT good!          |
|------------------------------------<steve <at> sxemacs.org>---|
issue-tracking | 30 Dec 22:21 2015

[Bug 177] New: horizontal scrolling doesn't work

Bug ID Summary Product Version Hardware OS Status Severity Priority Component Assignee Reporter QA Contact
177
horizontal scrolling doesn't work
SXEmacs
22.1.15
PC
Linux
NEW
minor
P5
General
steve <at> sxemacs.org
jay.p.belanger <at> gmail.com
sxemacs-devel <at> sxemacs.org

I'm using git SXEmacs. If I have a buffer with a line longer than the window and I wish to scroll left, typing `M-x scroll-left' will scroll left, show Command `scroll-left' is bound to keys: C-next, C-kp-next, C-x < in the echo area for a couple seconds, and then unscroll (return to the original point). This is the same whether truncate-lines is nil or non-nil. If I type `C-x <', then nothing at all happens. Similarly for `scroll-right' and `C-x >'.
You are receiving this mail because:
  • You are the QA Contact for the bug.

Gmane