Re: Broken espeakup on debian sid

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Actually Kirk I have not used pipewire yet so sorry, no clue.

I want Slint to stay fully accessible with speech, i.e. in consoles as in
graphical environments and switching back and forth, so I provide default
settings allowing that for all software involved. A user who don't like these settings may of course change them. I am not able to reproduce or even list all
scenarii and use cases, so can hardly give a sound advice. All I can say is
that so far no user complained that the settings din"t fit his or her scenario. But of course our user base several order of magnitudes smaller than that of big
distributions.

Cheers,
Didier

PS and OT: I test installing Slint in a nested VM (Qemu inside Windows in
Qemu VM in my Slint laptop). At some point in the installation I was asked a
password. espeakup in the VM inner didn't spell what I typed, but Laeticia
(Portuguese voice of RHVoice) did as Orca was running in the host. You are
never careful enough...

Le 18/09/2021 à 18:02, Kirk Reiser a écrit :
Thank you Didier: That was an excellent review. As much as I like not
building in pulseaudio support it would be nice to be able to include
pipewire support for folks that want it. Do you have any suggestions
about the difference between those two situations with respect to
espeakup?

Imo it would be nice if we could provide support for all scenarios,
mostly because the more complicated infrastructure becomes the more
difficult it is to debug problems and lots of folks just don't like
running all the bells and whistles. Personally I'll take lean over
fancy every time.

E.g. this particular problem appears to be a systems fuck up instead
of an actual problem with the code base.

   Kirk

On Sat, 18 Sep 2021, Didier Spaier wrote:

Hi Kirk and Janina:

In Slint at time of writing we have:
alsa-lib-1.2.5
espeakup-0.9.0
espeak-ng-git2b77dd27
linux-5.3.13
pcaudiolib-gita41d46e
sonic-gitba331411

No issue with espeakup, neither that I observed using it nor reported by users.
This makes me assume that the issue doesn't come from upstream.

FYI, Hhere are the binary dependencies, according to ldd:
dance[~]$ ldd /usr/bin/espeakup
    linux-vdso.so.1 (0x00007ffe7059d000)
    libespeak-ng.so.1 => /usr/lib64/libespeak-ng.so.1 (0x00007f284e5a5000)
    libasound.so.2 => /usr/lib64/libasound.so.2 (0x00007f284e37b000)
    libm.so.6 => /lib64/libm.so.6 (0x00007f284e072000)
    libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f284de55000)
    libc.so.6 => /lib64/libc.so.6 (0x00007f284da8c000)
    libpcaudio.so.0 => /usr/lib64/../lib64/libpcaudio.so.0 (0x00007f284e59c000)     libsonic.so.0 => /usr/lib64/../lib64/libsonic.so.0 (0x00007f284e596000)     libstdc++.so.6 => /usr/lib64/../lib64/libstdc++.so.6 (0x00007f284d710000)     libgcc_s.so.1 => /usr/lib64/../lib64/libgcc_s.so.1 (0x00007f284d4f9000)
    libdl.so.2 => /lib64/libdl.so.2 (0x00007f284d2f5000)
    librt.so.1 => /lib64/librt.so.1 (0x00007f284d0ed000)
    /lib64/ld-linux-x86-64.so.2 (0x00007f284e480000)

I don't know if that comes into play, but in Slint we don't redirect the sound
streams from alsa to pulseaudio, rather the opposite.

It could matter what I read in our ChangeLog at:
https://slackware.uk/slint/x86_64/slint-14.2.1/ChangeLog.txt
quoted below:
Tuesday 29 June 2021
pcaudiolib-gita41d46e-x86_64-2slint: Rebuilt
 Built without pulseaudio to avoid starting pulseaudio as root through
 espeakup. Thanks to Tony Seth for the heads-up.

so I have included these configure options for pcaudiolib:
--with-alsa
--without-pulseaudio

Cheers,
Didier

Le 18/09/2021 à 15:10, Janina Sajka a écrit :
Hi, Kirk:

I'm on Arch, not Debian, but I've had broken Espeakup ever since
alsa-lib-1.2.5 showed up. Since then Espeakup has been upversioned to
0.90, but this has not fixed things for me, though apparently it has for
Alexander now maintaning Espeakup.

I currently have two functional Linux machines, both fully upgraded as
of just about half an hour ago except as described below.

The newer hardware will run by hand with the command: 'espeakup -d', but
it's highly brittle and prone to crash the system so that a three finger
salute is needed.

If I want console access on this 2020 era box, I have to turn to fenrir,
which I'm just not as comfortable with--but that's another story.

My older machine is prevented by my /etc/pacman.conf from updating
alsa-lib and espeakup, and it runs just fine.

So, my suggestion is back off to the latest alsa-lib-1.2.4, and the
latest espeakup-0.8 you have, and you should be fine until someone
figures out what's really going on.

For anyone running Arch you want the following in your /etc/pacman.conf:

IgnorePkg   =espeakup
IgnorePkg   =alsa-lib

If you've already updated your Arch beyond these versions, you can
downgrade. You'll find a cache of previous versions in:

/var/cache/pacman/pkg/
Just cd on over there as root and check what you have with a command
like:

ls -1 alsa-lib* espeakup*

You can then downversion with pacman like this:

pacman -U ./[filename]

where [filename] is the full name you got from ls. I recommend using the
Speakup clipboard to get the command right.

Best,

Janina


Kirk Reiser writes:
Hi folks: Does anyone have espeakup running on debian sid with
libasound2 1.2.5? After I upgraded a few days ago I lost my speech
output. I have built espeakup and espeak-ng from the current repo on
them with no joy either.

Curious minds and all that type thing.

   Kirk







[Index of Archives]     [Linux for the Blind]     [Fedora Discussioin]     [Linux Kernel]     [Yosemite News]     [Big List of Linux Books]

  Powered by Linux