Joe Schmoe's questions

It looks like Python3.7 is in the repo (I can see it with apt-file search python3.7). But my main question is: Will it break anything?

From what I've read, it looks like the best option is to install "pyenv", but I don't know how complex that is or what it will entail.

Note that the underlying problem here is that I have an application in Python that now says that 3.6 is deprecated and it will need 3.7 or later in the future.

I've tried building Python from source (from python.org or whatever their main site is), but that hasn't worked. It builds OK, but then when I run my app, it says something or other isn't found. The problem seems to be that when Python is built (i.e., ./configure-d), it checks to see if certain things are already installed on your system (e.g., the various bits of "SSL" stuff) and if they aren't it just goes ahead and builds w/o telling you that stuff is missing. It's be better if it stopped and told you "You need to install such-and-such, then re-run ./configure")

I'm trying to get "gmplayer" to work. I did, eventually, get it to work, but it was a struggle, and I have a question (at the end of this post). Note: I first installed mplayer and it works just fine. That worked "out of the box". But when I installed mplayer-gui, then I had problems getting gmplayer working.

First, I was getting the error described at gmplayer can't run which boils down to doing (as root):

cd /usr/share/mplayer/skins/default
for FILE in *.png
do
    convert $FILE -define png:format=png24 $FILE
done

Note: convert is from ImageMagick.

But it still would not run. I get the following error message repeated infinitely until I CtrlC:

[vdpau] Error when calling vdp_device_create_x11: 1
Error opening/initializing the selected video_out (-vo) device.
Failed to open VDPAU backend libvdpau_i965.so: cannot open shared object file: No such file or directory

I then consult the man page and work out that I need to specify the video driver using the -vo switch. When I add: -vo xv to the gmplayer command line, it works (still generates a bunch of warnings/errors on the terminal screen, but it does work) and I am able to watch the video and use the GUI interface to control mplayer. Yey!

But now I am wondering where the libvpau_i965 library comes from and if I can install it to make the error go away (i.e., so that it will work without -vo xv).

I do:

apt-file search libvpau_i965.so

and it doesn't find anything. So, that file doesn't seem to be in any of my sources for apt-getting. Any idea as to where to find it?

Edit to add:

I've also noticed that there is no command line or config file option to set the playback volume. gmplayer always defaults to (i.e., starts up as) zero volume, which is annoying. mplayer usually defaults to whatever the volume last was.

Anyway, here is a workaround script for gmplayer to set the volume up on startup:

#!/bin/bash
[ -r "$1" ] || exec echo "Not found: $1"
{ sleep 3;xdotool type '***********************************************'; } &
gmplayer "$1"

Bionic Beaver, 18.04.

A funny thing happened today. I have a system, working fine, booted up to the desktop - uptime about 80 days, lots of windows open on the desktop. Everything fine, until... GUI is running on tty1 (as expected).

In trying to debug another problem, I decide to switch away from the desktop GUI to a virtual console and login. So,I do "sudo chvt 2" and get to the login prompt on tty2. So far, so good. I log in as myself (on tty2), do a few things, then logout (via ^D) - expecting to get back to a login prompt, from which I can then switch back to my GUI desktop session.

However, next thing that happens is that I am presented with a pure purple screen, then about 10 seconds later, a GUI screen asking for my password. I log in and am presented with a clean, fresh desktop. All windows that had previously been there are gone (meaning, I have to figure out what was there before, and restart everything).

It looks like somehow, logging out of tty2, sent a Hangup signal to all the processes running on the GUI desktop, including the desktop itself, forcing you to log in again and restarts whatever had been running before.

Why does this happen?

Some additional information:

1. After the desktop re-start - and after I've run my scripts to open up my usual windows (programs) - I find that there are a maze of processes running on tty1, running as user "gdm", and also a maze of processed on tty3, running as user "me". All of these processes (on both tty1 and tty3) show as being started at the time of the desktop re-start. There is a single process (a getty) running on tty2.

2. There is a single file in /var/crash, timestamped at the time of the desktop re-start, that corresponds to one of my processes (i.e., something I had running in one of my terminal windows at the time of the crash). That file shows it to have been killed by signal 6 (SIGABRT). SIGABRT is usually caused by a program failing an assert(3).

Every so often - I think it is whenever my ISP "glitches" (i.e., when there is a brief loss of Internet connectivity), my Ubuntu system screen gets covered with a glossy film (bluish-grey) that says something about not being able to connect to something or other.

I have determined that this is caused by a program (process) called "gnome-shell-portal-helper" and if I kill that process, the glossy film goes away.

How to turn this off permanently?

Running Ubuntu 18.04 on x64.

I need to COMPLETELY turn off the "automatic updates reminder" stuff on my Ubuntu system. The problem is that the machine is doing some unattended processing and the updates reminders interfere with it.

A while back, I Googled for help on this and got some pointers to go into Software Updater "settings" and set the "Remind me when" thing to "Never". That sorta (but not really) worked. It disables the big screen that comes up (the one that tells you that "We have new software for you"), but not the little one.

Anyway, I could go on more about how that doesn't work, but the point is that it is clear that it really, really wants you to do this and they have made the config hard to work - to really turn it off.

So, I need something lower level - not so "user-y" or "GUI-I" - because that just doesn't work. There has to be something lower level - some service or something that you can kill or disable to make this go away. Can you help me, please?

Running Ubuntu 18.04 on x64.

For one particular package, namely Vim/Gvim, there are a few custom files that I install in one of the directories (specifically, the "syntax" directory). I have found that when I do the usual update/upgrade thing, and a new version of the package gets installed, it over-writes my custom file.

Took me a while to figure out why my syntax highlighting was no longer working as before. Would (obviously) like to take some steps to make sure this doesn't happen again the next time the package is upgraded.

Seems to me to boil down to 3 possibilities:

1. Make the file in question immutable (with chattr). This is the first thing I thought of and is, alas, brute force, but the question then becomes: Would this cause the upgrade process to get confused and space out? (Note; This is the question I really want to hear from the community about)

2. Is there some way in the apt/apt-get/etc ecosystem to mark a file as not-to-be-changed? Not a whole package, mind you (I know how to do that with apt-mark), but just one single file.

3. Given that we're talking about Vim, I think it is actually possible to put local versions of the syntax files in your home directory, under ~/.vim - but the details escape me at the moment and I'd rather not do it this way if possible.

I am running Ubuntu 18.04 on x64.

The machine has been up for a week or so, doing all the usual things it does, until today, I noticed that it had dropped off my local network. At the console, the "ifconfig" commands shows no IP address assigned to the eth0 device (Note: It's not actually eth0 anymore; they now give it a name which is a random collection of letters, but the idea is the same. I still think of it as eth0).

Now, as to the cause of this; I don't really care; I just assume some kind of software glitch. The question is: How to fix? I would imagein that the easiest solution is just to reboot, but I'd prefer to avoid that.

In the olden days, you could just unplug/replug the ethernet cable and it would fix itself. This always works under Windows, for example. It always worked in Linux, up until recently. I've noticed that current/later-day versions of Linux don't seem to fix themselves like they used to.

Further, on some version of Linux (but not Ubuntu, apparently), you could figure out which DHCP program (e.g., dhcpcd) they were using (it seems they keep changing it), usually by searching all running processes for a process containing the string "dh" and then you could figure out from the man pages how to get it running again. However, that method seems not to work in Ubuntu 18.04. In any case, I could not figure out which program was doing the dhcp'ing.

Anyway, I was able to get it working again by doing the following manual stuff (Note: This works because I happened to know the IP address that it should and did have before):

1. ifconfig eth0-but-not really w.x.y.z
2. route add default gw w.x.y.1
3. edit /etc/resolv.conf - put in the nameserver (On my network, also w.x.y.1)

But, obviously, I'd like to know the right way to do it, to get the router to assign the IP address like it is supposed to.

Note: I looked through some of the suggested "similar questions" and didn't find anything that looked conclusive. Plus, it seems most of them are 6 years old (from 2014), so I'm hoping for something more up-to-date (and more likely to "just work").

I have a 64 bit Ubuntu system that works fine. I would like to be able to build a 32 bit version of, say, "hello, world". This is mostly an academic pursuit, but it would be convenient to get it working. It would be nice if compiling with "-m32" would "just work", but it doesn't. Worse, my memory is that this used to "just work" (in an older version of 64 bit Linux), but no longer works.

Observe:

$ cat hello.c
#include <stdio.h>

int main() { puts("hello, world"); return 0; }
$ gcc -m32 hello.c
In file included from hello.c:1:0:
/usr/include/stdio.h:27:10: fatal error: bits/libc-header-start.h: No such file or directory
 #include <bits/libc-header-start.h>
          ^~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
$

A bit of poking around suggests that installing these packages (see below) might help, so I did:

apt-get install libc6-dev-i386-x32-cross

which installed the following:

The following NEW packages will be installed:


libc6-dev-i386-x32-cross libc6-dev-x32-cross libc6-i386-x32-cross
  libc6-x32-cross linux-libc-dev-x32-cross

After that, with a little bit of fudging, I was able to get it to compile, but not link. The link phase gives these error msgs:

/usr/bin/ld: cannot find Scrt1.o: No such file or directory
/usr/bin/ld: cannot find crti.o: No such file or directory
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/7/libgcc.a when searching for -lgcc
/usr/bin/ld: cannot find -lgcc
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/7/libgcc.a when searching for -lgcc
/usr/bin/ld: cannot find -lgcc
collect2: error: ld returned 1 exit status

And I can't get any farther than that.

So, any advice will be appreciated.

First note: There seem to be several existing threads on this general subject. However, I looked them over and they all seem to be of the "unmet dependencies"/can't install variety. My problem is different; it installs; but fails to run.

I did the usual apt-get update; apt-get install lighttpd and during the install, the following error was displayed onscreen:

/usr/sbin/lighttpd: Symbol `FamErrlist' has different size in shared object, consider re-linking
/usr/sbin/lighttpd: symbol lookup error: /usr/sbin/lighttpd: undefined symbol: FAMNoExists

Along with a message to the effect that lighttpd could not be started.

I've done a fair amount of other research trying to get this working, all to no avail. Note: I even downloaded a .deb on a different Ubuntu system, where the binary does work, but when I take the .deb and unpack it on the 18.04 system, it fails. I've tried to compile the source, but the source package is weird; it does not have a ./configure script (just configure.ac, which I tried, unsuccessfully, to convert to "configure" using "autoconf")

Anyway, I ended up installing something called "h2static", which seems to be a pretty good replacement, but I'm still curious why "lighttpd" doesn't work.

Update: Somebody requested further information. Note that I have long since uninstalled the package, so anything related to its running as an installed package is gone, but here is the "ldd" output:

linux-vdso.so.1 (0x00007ffcc8f43000)
        libpcre.so.3 => /lib/x86_64-linux-gnu/libpcre.so.3 (0x00007fbd9246a000)
        libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007fbd92266000)
        libattr.so.1 => /lib/x86_64-linux-gnu/libattr.so.1 (0x00007fbd92061000)
        libssl.so.1.1 => /usr/lib/x86_64-linux-gnu/libssl.so.1.1 (0x00007fbd91dd4000)
        libcrypto.so.1.1 => /usr/lib/x86_64-linux-gnu/libcrypto.so.1.1 (0x00007fbd91909000)
        libfam.so.0 => /usr/lib/x86_64-linux-gnu/libfam.so.0 (0x00007fbd91700000)
        libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007fbd9130f000)
        libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007fbd910f0000)
        /lib64/ld-linux-x86-64.so.2 (0x00007fbd9291b000)
        libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007fbd90d67000)
        libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007fbd90b4f000)
        libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007fbd907b1000)

And here is the error message that any attempt to run the binary generates (This is the same as posted above, but here it is again for clarity):

/usr/sbin/lighttpd: Symbol `FamErrlist' has different size in shared object, consider re-linking
/usr/sbin/lighttpd: symbol lookup error: usr/sbin/lighttpd: undefined symbol: FAMNoExists

I have used the virtual consoles (tty1-6) for decades in various Linuxes. I recently installed Ubuntu 18.04 (Bionic Beaver) on an x86-64 box. Everything is fine, except that the VCs don't work. As far as I can tell, it is by design that people are now only supposed to use the GUI (like in Windows). I would like to be able to use the VCs.

Note:
I've noticed (from ps output) that a lot of the GUI processes are running on tty1, so I assume that the GUI gets launched from there).

What I've tried:

1. Ctrl-Alt-F1 (or F2 or F3 or ...). Nothing happens. By "nothing happens", I mean, literally, nothing happens.
   Note:
   I am familiar with one system where you can switch to the virtual consoles (from the GUI), and it really does switch, but the screen stays on the GUI. You can now type and you are typing on the virtual console, but the screen makes it look like you are still on the GUI. I mention all this to say that, no, this is not the case here. You are still typing into the GUI.
2. (as root) chvt 1. Nothing happens. Prompt comes right back and nothing has changed.
3. (as root) chvt 2. Process hangs until you hit Ctrl/C. Weird...!

4. (as root) strace chvt 2. Lots of strace output, ending with something like:

   ioctl(3,VT_WAITACTIVE ...)
   

and then it hangs (until you hit CTRL-C).
Next, following some advice I found on this board, I tried editing the file:

/etc/systemd/logind.conf

and uncommenting the first two active lines, which are:

NAutoVTs=6
ReserveVT=6

I tried first just uncommenting the first one, then rebooted, nothing changed.
So I then uncommented the second one as well, rebooted, again, no change.

So, what to do?
What is next?