(The same code is in upstream svn trunk.)

Consider this:

$ python
Python 2.7.16 (default, Oct 10 2019, 22:02:15)
[GCC 8.3.0] on linux2
Type "help", "copyright", "credits" or "license" for more information.
False
$ cat /proc/1/cmdline
/sbin/init$ ls -l /sbin/init
lrwxrwxrwx 1 root root 20 Apr 27  2020 /sbin/init -> /lib/systemd/systemd*

The test is confused by the symlink and returns a false negative.

I had been wondering why systemd-run wasn't being used even if I explicitly asked for it.

Fix with caution, as doing so may expose latent bugs.

Thanks for report. However this version is obsolete and the problem is most
certainly fixed in newer release. I recommend to upgrade to version from
"buster-backports" as I hope it may have the fix already...

---

The end cannot justify the means for the simple and obvious reason that the
means employed determine the nature of the ends produced.
    -- Aldous Huxley

---

And how long a lockdown is enough? If we open now, will lockdown recur in
autumn? Next year? Whenever authoritarianism so wishes? No dictatorship
could imagine a better precedent for absolute control.
    -- https://www.bmj.com/content/369/bmj.m1924.long
    :: BMJ 2020;369:m1924 "Should governments continue lockdown to slow the spread of covid-19?"

control: found -1 3.0.4+dfsg1-1~bpo10+1

As I mentioned parenthetically in the original report, the same flawed
detection code is on upstream svn trunk. I've also checked that it is
still called from xpra/scripts/main.py in a relevant way. As far as I can
tell it affects all versions so far. I have explicitly checked 3.0.4:

$ python3
Python 3.7.3 (default, Jul 25 2020, 13:03:44)
[GCC 8.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
False
xpra v3.0.4-r24778

I now also have a suggested fix: read /proc/1/status and look for a line
matching ^Name:[[:space:]]+systemd$ (or whatever the proper regexp syntax is)
instead of /proc/1/cmdline which may read /sbin/init even on systemd.