Are you sure the latest-skype.sh script will download and package different versions when setting the $VERSION variable?

The script uses $VERSION for the destination directory, and for the final package name. But the wget link used in the script is always using the "latest" rpm.

I just tried running the script twice (once with VERSION=8.29.0.50, and once without the VERSION). The size of the rpm downloaded was exactly the same for both times. And with both packages skypeforlinux fails to work.

I so wish something else besides Skype did all that Skype can do. Every time Skype misbehaves I try to convince family and friends to try alternatives and they don't want to or they have an Apple product and that alternative is unsuitable (sigh)

Same problem here <sigh>.


I tried the 8.30.76.41 version (as root too), as Eric reported already, same error.

I use the rpm version as it's easier to check version number before downloading. My script checks the for the latest version from https://repo.skype.com/latest/, it downloads from https://repo.skype.com/rpm/stable/

The 8.30.76.41 version is listed skypeforlinux-64-insider.rpm in /latest/ and found in https://repo.skype.com/rpm/unstable/.

Back to the latest version I have installed 8.18.0.6. Check that. I noted that 8.29.0.50 works, so upgraded to that version. I does indeed work.

1 members found this post helpful.

As a further datum point, I have gnome-3.28 installed on slackware-current and when I want to run gnome I boot up with systemd (which I also have installed). When booting slackware-current under systemd, skype version 8.30.76.41 (obtained from the repository mentioned in comment #3) works for me.

However if I boot up under SysV init then skype fails to start. According to /var/log/messages dbus cannot activate the org.freedesktop.login1 service and it is skype which makes the dbus request for that service. I am pretty certain that that is the service provided by systemd's logind.

So it may well be that skype now requires logind. If so, that dependency may well be resolved by users installing elogind, which is a bit like eudev - ripped out of systemd and separately maintained, in this case for the purposes of the GUIX distribution; possibly gentoo also uses it, I am not sure. (If slackware were to adopt elogind it could probably dispense with ConsoleKit.)

As it happens there is an entirely different problem with skype version 8.30.0.50 - that fails to find my microphone. That issue is resolved in version 8.30.76.41.

Edit: 1) I can confirm that org.freedesktop.login1 is owned by logind. 2) Bummer: elogind requires PAM, which real slackers probably wouldn't like. (Still, if you want to be a real slacker I suppose you have to make sacrifices.)

1 members found this post helpful.

Good catch! I didn't look close enough to realize the second wget was hardcoded to the "latest" url. I don't have the ability to generate a diff right now (work computer running Windows and I couldn't find any websites that would generate one for you), but if you change the following line (line number 145):

with this:

it should work with any versions they have in their repo.

I still have systemd on my laptop (at one point was running Dlackware, removed it, but I think I will need to do a fresh install to get rid of systemd), so I can give it a try later today as a test just to confirm.

Are you sure? Because their README states that PAM is an optional dependency.

https://github.com/elogind/elogind

Also, ktown has elogind as a dependency and it doesn't require PAM... but it is an older version of elogind (219.14 vs 238.1), so maybe it's been added as a required dependency more recently, but without updating the README.

You are right and I am wrong. So not such a bummer after all. And 'objdump -p /usr/share/skypeforlinux/skypeforlinux' doesn't show any unwanted linker dependencies, as far as I can see.

Although installing PAM wouldn't be a bad idea in my view.

If that's the case, I wonder if installing elogind solves the problem with skypeforlinux.

This could cause the thread to devolve into something unrelated to skypeforlinux, so hopefully people leave it be

You don't need to install dlackware just to boot up slackware-current with systemd. dlackware is too intrusive in my view, and anyway is a bit behind slackware-current. Perhaps surprisingly, there are only four or five slackware packages you need to recompile to make them systemd-aware and two or three for PAM (if you are interested I can look up what they are); the main effort is in writing systemd unit files to mirror whatever you would start up with SysVinit via /etc/rc.d or with crond or atd - I have a little over a dozen of them. systemd and SysV init run surprising well when installed together: you just choose the init to run at boot time.

It quite possibly does. Give it a go.

I don't have my computer set up right now (I've been in the process of moving the last several months and we're finally ready to get our stuff moved into the new house... hopefully next week), so I don't have the ability to test this right now.

OK, let me very quickly quell this bit of mis-information.
Ktown does not have elogind as a dependency.

What you point to is the source of an elogind package for my December 2017 release of ktown_testing. There has never been an elogind in the "latest" repository of ktown. The elogind package was once used in an attempt to create a working Wayland desktop. I dropped elogind in favor of ConsoleKit2 after that implemented the required parts of the Logind API.
The December 2017 release of ktown is available only for reference purposes. It is the last release targeting Slackware 14.2 (still supporting the old KDE4 packages) but ktown has since moved on for another 9 months and targeting -current only.

Is that source archive just for 14.2? Is there not one for -current? The elogind source is available in each version (5, 5_17.11, latest, testing). If it isn't being used, that could lead others to come to the same, incorrect conclusions that I did.

Anyway, thanks for the heads up!

People should inform themselves better then. The packages are all that counts.
There's more sources in there that are not being used for the "testing" branch.

I use a single source tree for everything and build what's relevant, skip what's not.