Using xonsh shell with pyenv on Ubuntu 18.04 – and a few errors I had getting source-bash to work 😉

When I heard about the xonsh project – https://xon.sh I thought it sounded great, as I really enjoy the python language and am much more comfortable using python to write scripts than I am using bash. I also find the bash syntax very confusing, and am never sure of the right way to write a script or use a variable.

So, I decided to install it and give it a try! When using python on Linux I always use the pyenv tool to install and manage my python versions, as well as my virtualenvs. This tool allows me to easily install the latest version of python, as well as create a virtualenv for a specific project and keep all the packages that project uses separate from each other and my system python. It’s a great tool, and I can highly recommend using it.

pyenv installs all python versions into your user profile, and allows you to select from them on the command line using

pyenv shell 3.7.0

To get all the pyenv commands, along with the pyenv command completion (which is very good), you add an init command to the end of your .bashrc file. This is how it looks in mine:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

Excellent! When running in bash this sets up pyenv, then sets my shell to use python 3.7.0. So let’s install xonsh:

jtuckey@jay-x1-carbon:~$ pip install --user xonsh
Collecting xonsh
  Using cached https://files.pythonhosted.org/packages/58/16/fce8ecc880a44dfcb06b22f638df0b6982ad588eb0f2080fbd5218a42340/xonsh-0.8.0.tar.gz
Installing collected packages: xonsh
  Running setup.py install for xonsh ... done
Successfully installed xonsh-0.8.0

Looking good so far! I can now use the xonsh command to hop into a xonsh shell, and then use all the goodness of a python-based shell:

jtuckey@jay-x1-carbon:~$ xonsh
jtuckey@jay-x1-carbon ~ $ ${'XONSH_VERSION'}
'0.8.0'
jtuckey@jay-x1-carbon ~ $ history info
backend: json
sessionid: 30e7e6d4-4945-4f99-9b65-d38a2ab61da2
filename: /home/jtuckey/.local/share/xonsh/xonsh-30e7e6d4-4945-4f99-9b65-d38a2ab61da2.json
length: 1
buffersize: 100
bufferlength: 1
gc options: (8128, 'commands')

Now that I have it installed and am getting all the good features, I naturally wanted it as my default shell. My first attempt at this was to simply put it the end of my .bashrc file:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

xonsh

This works! However, to quit the shell, I need to exit twice:

jtuckey@jay-x1-carbon ~ <xonsh>$ exit
jtuckey@jay-x1-carbon:~$ exit

Ok, maybe let’s put it in an auto-exiting block. Back to the .bashrc:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

# Start xonsh shell
if xonsh; then
  exit
fi

This works even better. Now it will automatically exit when I’m done using xonsh, but if it can’t find xonsh it will stay in bash. However, with this setup, I try to do some work, which involves activating one of my pyenv virtualenvs. This gives me an error:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.

Ok, so clearly the pyenv setup chunk from bash isn’t carrying across nicely from bash to xonsh. The path looks alright:

jtuckey@jay-x1-carbon ~ <xonsh>$ $PATH
\EnvPath(
['/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims',
 '/home/jtuckey/.pyenv/shims',
 '/home/jtuckey/.pyenv/bin',
 '/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims',
 '/home/jtuckey/.pyenv/shims',
 '/home/jtuckey/.pyenv/bin',
 '/home/jtuckey/.local/bin',
 '/usr/local/sbin',
 '/usr/local/bin',
 '/usr/sbin',
 '/usr/bin',
 '/sbin',
 '/bin',
 '/usr/games',
 '/usr/local/games',
 '/snap/bin']
)

Let’s have a look at what those two pyenv init commands. What they are doing is generating a bit of code using pyenv init - pyenv virtualenv-init - and then evaluating it in the local scope. We can easily see the code that’s being generated:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv init -
export PATH="/home/jtuckey/.pyenv/shims:${PATH}"
export PYENV_SHELL=python3.7
command pyenv rehash 2>/dev/null
pyenv() {
  local command
  command="${1:-}"
  if [ "$#" -gt 0 ]; then
    shift
  fi

  case "$command" in
  activate|deactivate|rehash|shell)
    eval "$(pyenv "sh-$command" "$@")";;
  *)
    command pyenv "$command" "$@";;
  esac
}
jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv virtualenv-init -
export PATH="/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims:${PATH}";
export PYENV_VIRTUALENV_INIT=1;
_pyenv_virtualenv_hook() {
  local ret=$?
  if [ -n "$VIRTUAL_ENV" ]; then
    eval "$(pyenv sh-activate --quiet || pyenv sh-deactivate --quiet || true)" || true
  else
    eval "$(pyenv sh-activate --quiet || true)" || true
  fi
  return $ret
};
if ! [[ "$PROMPT_COMMAND" =~ _pyenv_virtualenv_hook ]]; then
  PROMPT_COMMAND="_pyenv_virtualenv_hook;$PROMPT_COMMAND";
fi

Ok, so if we can load these as bash, we can get all the commands correctly into xonsh. Fortunately there is a command in xonsh just for this:

jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash --help
usage: source-foreign [-h] [-i INTERACTIVE] [-l LOGIN] [--envcmd ENVCMD]
                      [--aliascmd ALIASCMD] [--extra-args EXTRA_ARGS]
                      [-s SAFE] [-p PREVCMD] [--postcmd POSTCMD]
                      [--funcscmd FUNCSCMD] [--sourcer SOURCER]
                      [--use-tmpfile USE_TMPFILE]
                      [--seterrprevcmd SETERRPREVCMD]
                      [--seterrpostcmd SETERRPOSTCMD] [--overwrite-aliases]
                      [--suppress-skip-message] [--show] [-d]
                      shell files_or_code [files_or_code ...]

Sources a file written in a foreign shell language.

positional arguments:
  shell                 Name or path to the foreign shell
  files_or_code         file paths to source or code in the target language.
...

jtuckey@jay-x1-carbon ~ <xonsh>$ aliases
xonsh.aliases.Aliases(
{...
 'source-bash': ['source-foreign', 'bash', '--sourcer=source']
...}

So we should be able to just source the output of the pyenv commands. Let’s try it:

jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv init - bash)
# Shell just hangs at this point....

Sooo…. that’s not working. My shell just locks up. I can’t even Ctrl-C to kill it.

What’s going on here? After a bit of investigation, what seems to be happening is this: when you use source-bash the bash shell launched to run the source is running the .bashrc file, which puts it back into xonsh, where it gets stuck. This makes sense, although it’s hard to work out. To test this, lets get rid of the xonsh lines from .bashrc and see if it works correctly then.

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

# Start xonsh shell
#if xonsh; then
#  exit
#fi

And try running it manually

jtuckey@jay-x1-carbon:~$ xonsh
jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv init - bash)
Skipping application of 'll' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
Skipping application of 'ls' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
jtuckey@jay-x1-carbon ~ <xonsh>$ 

Looking better. Now I’m back where I started, though, so how can I run xonsh as my shell. From the xonsh tutorial there is this line:

Alternatively, you can setup your terminal emulator (xterm, gnome-terminal, etc) to run xonsh automatically when it starts up. This is recommended.

However, in my gnome-terminal if I try setting xonsh as the command it can’t find xonsh. This is because the python 3.7 environment where xonsh lives is initialised during bash startup, not gui startup, so my gnome-terminal can’t find the xonsh command.

Ok, so how do I put something into my whole user environment? That’s what the .profile file in your user profile is for. Let’s put the pyenv init stuff into the end of my .profile file:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

Now with a logoff and back on, let’s see what happens. Now when I set gnome-terminal to run xonsh, it works. However:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.


jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv virtualenv-init - bash)
Skipping application of 'll' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
Skipping application of 'ls' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.

Hmmm, still not quite there, but closer.

When I heard about the xonsh project – https://xon.sh I thought it sounded great, as I really enjoy the python language and am much more comfortable using python to write scripts than I am using bash. I also find the bash syntax very confusing, and am never sure of the right way to write a script or use a variable.

So, I decided to install it and give it a try! When using python on Linux I always use the pyenv tool to install and manage my python versions, as well as my virtualenvs. This tool allows me to easily install the latest version of python, as well as create a virtualenv for a specific project and keep all the packages that project uses separate from each other and my system python. It’s a great tool, and I can highly recommend using it.

pyenv installs all python versions into your user profile, and allows you to select from them on the command line using

pyenv shell 3.7.0

To get all the pyenv commands, along with the pyenv command completion (which is very good), you add an init command to the end of your .bashrc file. This is how it looks in mine:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

Excellent! When running in bash this sets up pyenv, then sets my shell to use python 3.7.0. So let’s install xonsh:

jtuckey@jay-x1-carbon:~$ pip install --user xonsh
Collecting xonsh
  Using cached https://files.pythonhosted.org/packages/58/16/fce8ecc880a44dfcb06b22f638df0b6982ad588eb0f2080fbd5218a42340/xonsh-0.8.0.tar.gz
Installing collected packages: xonsh
  Running setup.py install for xonsh ... done
Successfully installed xonsh-0.8.0

Looking good so far! I can now use the xonsh command to hop into a xonsh shell, and then use all the goodness of a python-based shell:

jtuckey@jay-x1-carbon:~$ xonsh
jtuckey@jay-x1-carbon ~ $ ${'XONSH_VERSION'}
'0.8.0'
jtuckey@jay-x1-carbon ~ $ history info
backend: json
sessionid: 30e7e6d4-4945-4f99-9b65-d38a2ab61da2
filename: /home/jtuckey/.local/share/xonsh/xonsh-30e7e6d4-4945-4f99-9b65-d38a2ab61da2.json
length: 1
buffersize: 100
bufferlength: 1
gc options: (8128, 'commands')

Now that I have it installed and am getting all the good features, I naturally wanted it as my default shell. My first attempt at this was to simply put it the end of my .bashrc file:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

xonsh

This works! However, to quit the shell, I need to exit twice:

jtuckey@jay-x1-carbon ~ <xonsh>$ exit
jtuckey@jay-x1-carbon:~$ exit

Ok, maybe let’s put it in an auto-exiting block. Back to the .bashrc:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

# Start xonsh shell
if xonsh; then
  exit
fi

This works even better. Now it will automatically exit when I’m done using xonsh, but if it can’t find xonsh it will stay in bash. However, with this setup, I try to do some work, which involves activating one of my pyenv virtualenvs. This gives me an error:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.

Ok, so clearly the pyenv setup chunk from bash isn’t carrying across nicely from bash to xonsh. The path looks alright:

jtuckey@jay-x1-carbon ~ <xonsh>$ $PATH
\EnvPath(
['/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims',
 '/home/jtuckey/.pyenv/shims',
 '/home/jtuckey/.pyenv/bin',
 '/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims',
 '/home/jtuckey/.pyenv/shims',
 '/home/jtuckey/.pyenv/bin',
 '/home/jtuckey/.local/bin',
 '/usr/local/sbin',
 '/usr/local/bin',
 '/usr/sbin',
 '/usr/bin',
 '/sbin',
 '/bin',
 '/usr/games',
 '/usr/local/games',
 '/snap/bin']
)

Let’s have a look at what those two pyenv init commands. What they are doing is generating a bit of code using pyenv init - pyenv virtualenv-init - and then evaluating it in the local scope. We can easily see the code that’s being generated:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv init -
export PATH="/home/jtuckey/.pyenv/shims:${PATH}"
export PYENV_SHELL=python3.7
command pyenv rehash 2>/dev/null
pyenv() {
  local command
  command="${1:-}"
  if [ "$#" -gt 0 ]; then
    shift
  fi

  case "$command" in
  activate|deactivate|rehash|shell)
    eval "$(pyenv "sh-$command" "$@")";;
  *)
    command pyenv "$command" "$@";;
  esac
}
jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv virtualenv-init -
export PATH="/home/jtuckey/.pyenv/plugins/pyenv-virtualenv/shims:${PATH}";
export PYENV_VIRTUALENV_INIT=1;
_pyenv_virtualenv_hook() {
  local ret=$?
  if [ -n "$VIRTUAL_ENV" ]; then
    eval "$(pyenv sh-activate --quiet || pyenv sh-deactivate --quiet || true)" || true
  else
    eval "$(pyenv sh-activate --quiet || true)" || true
  fi
  return $ret
};
if ! [[ "$PROMPT_COMMAND" =~ _pyenv_virtualenv_hook ]]; then
  PROMPT_COMMAND="_pyenv_virtualenv_hook;$PROMPT_COMMAND";
fi

Ok, so if we can load these as bash, we can get all the commands correctly into xonsh. Fortunately there is a command in xonsh just for this:

jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash --help
usage: source-foreign [-h] [-i INTERACTIVE] [-l LOGIN] [--envcmd ENVCMD]
                      [--aliascmd ALIASCMD] [--extra-args EXTRA_ARGS]
                      [-s SAFE] [-p PREVCMD] [--postcmd POSTCMD]
                      [--funcscmd FUNCSCMD] [--sourcer SOURCER]
                      [--use-tmpfile USE_TMPFILE]
                      [--seterrprevcmd SETERRPREVCMD]
                      [--seterrpostcmd SETERRPOSTCMD] [--overwrite-aliases]
                      [--suppress-skip-message] [--show] [-d]
                      shell files_or_code [files_or_code ...]

Sources a file written in a foreign shell language.

positional arguments:
  shell                 Name or path to the foreign shell
  files_or_code         file paths to source or code in the target language.
...

jtuckey@jay-x1-carbon ~ <xonsh>$ aliases
xonsh.aliases.Aliases(
{...
 'source-bash': ['source-foreign', 'bash', '--sourcer=source']
...}

So we should be able to just source the output of the pyenv commands. Let’s try it:

jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv init - bash)
# Shell just hangs at this point....

Sooo…. that’s not working. My shell just locks up. I can’t even Ctrl-C to kill it.

What’s going on here? After a bit of investigation, what seems to be happening is this: when you use source-bash the bash shell launched to run the source is running the .bashrc file, which puts it back into xonsh, where it gets stuck. This makes sense, although it’s hard to work out. To test this, lets get rid of the xonsh lines from .bashrc and see if it works correctly then.

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

# Start xonsh shell
#if xonsh; then
#  exit
#fi

And try running it manually

jtuckey@jay-x1-carbon:~$ xonsh
jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv init - bash)
Skipping application of 'll' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
Skipping application of 'ls' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
jtuckey@jay-x1-carbon ~ <xonsh>$ 

Looking better. Now I’m back where I started, though, so how can I run xonsh as my shell. From the xonsh tutorial there is this line:

Alternatively, you can setup your terminal emulator (xterm, gnome-terminal, etc) to run xonsh automatically when it starts up. This is recommended.

However, in my gnome-terminal if I try setting xonsh as the command it can’t find xonsh. This is because the python 3.7 environment where xonsh lives is initialised during bash startup, not gui startup, so my gnome-terminal can’t find the xonsh command.

Ok, so how do I put something into my whole user environment? That’s what the .profile file in your user profile is for. Let’s put the pyenv init stuff into the end of my .profile file:

# Load pyenv automatically by adding
# the following to ~/.bash_profile:

export PATH="/home/jtuckey/.pyenv/bin:$PATH"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

pyenv shell 3.7.0

Now with a logoff and back on, let’s see what happens. Now when I set gnome-terminal to run xonsh, it works. However:

jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.


jtuckey@jay-x1-carbon ~ <xonsh>$ source-bash $(pyenv virtualenv-init - bash)
Skipping application of 'll' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
Skipping application of 'ls' alias from 'bash' since it shares a name with an existing xonsh alias. Use "--overwrite-alias" option to apply it anyway.You may prevent this message with "--suppress-skip-message" or "$FOREIGN_ALIASES_SUPPRESS_SKIP_MESSAGE = True".
jtuckey@jay-x1-carbon ~ <xonsh>$ pyenv activate godev

Failed to activate virtualenv.

Perhaps pyenv-virtualenv has not been loaded into your shell properly.
Please restart current shell and try again.

Hmmm, still not quite there, but closer. After some further research, I found this on a page about virtual environments and xonsh.

The usual tools for creating Python virtual environments—venv, virtualenv, pew—don’t play well with xonsh. …. Luckily, xonsh ships with its own virtual environments manager called Vox.

https://xon.sh/python_virtual_environments.html

Sounds like vox is the way to go. To load vox you just need to load it. However, by default it doesn’t look for virtualenvs in the pyenv path. This is easily fixable by setting the $VIRTUALENV_HOME variable:

jay@jay-alienware-ubuntu:~:x$ xontrib load vox                                                                               
jay@jay-alienware-ubuntu:~:x$ vox list                                                                                       
No environments available. Create one with "vox new".

jay@jay-alienware-ubuntu:~:x$ $VIRTUALENV_HOME = $HOME +'/.pyenv/versions'                                                   
jay@jay-alienware-ubuntu:~:x$ vox list                                                                                       
Available environments:
3.6.2/envs/test-install
3.6.2/envs/tests
3.6.4/envs/mdb
3.7.0/envs/xonshdev
jay@jay-alienware-ubuntu:~:x$ vox activate tests                                                                             
Activated "tests".

(tests) jay@jay-alienware-ubuntu:~:x$ vox deactivate                                                                         
Deactivated "tests".

jay@jay-alienware-ubuntu:~:x$  

Be aware that creating a virtualenv through vox will just use your currently set python version from pyenv. However, you can still use pyenv to create a new virtualenv of any version, and then use vox to activate it:

jay@jay-alienware-ubuntu:~:x$ pyenv virtualenv 3.6.4 newenv                                                                  
Requirement already satisfied: setuptools in /home/jay/.pyenv/versions/3.6.4/envs/newenv/lib/python3.6/site-packages
Requirement already satisfied: pip in /home/jay/.pyenv/versions/3.6.4/envs/newenv/lib/python3.6/site-packages
jay@jay-alienware-ubuntu:~:x$ vox activate newenv                                                                            
Activated "newenv".

(newenv) jay@jay-alienware-ubuntu:~:x$  

So I now have a fully working pyenv+xonsh setup! The only thing that’s missing is proper tab-completion of the pyenv commands, which is something I may look into in the future, but for now I’m able to do my work easily without leaving the xonsh shell, which is great!

Fix for Windows Server 2016 Stuck at “Downloading 0%” from WSUS server – UpdateServiceUrlAlternate Registry Key

I ran into an issue recently with our Windows Server 2016 servers not installing their updates. Instead, all our 2016 servers would get stuck with the message “Updates are available. Downloading updates 0%” however they would never actually download the updates. These servers were configured to our WSUS server, which was also a Server 2016 machine.

The first thing I investigated was to look at the logs on the client server. I tried using the Get-WindowsUpdateLog cmdlet in Powershell, however, the log that was generated was useless because every line had no date and no format information (what the hell, Microsoft!). The logs look like this:

1601/01/01 09:30:00.0000000 872   1104                  Unknown( 39): GUID=50a5ef43-b2b6-3c52-145b-a386d8769665 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 38): GUID=50a5ef43-b2b6-3c52-145b-a386d8769665 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 34): GUID=7b9bf239-47b9-3688-3a9e-14f09f262608 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 39): GUID=50a5ef43-b2b6-3c52-145b-a386d8769665 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 38): GUID=50a5ef43-b2b6-3c52-145b-a386d8769665 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 34): GUID=7b9bf239-47b9-3688-3a9e-14f09f262608 (No Format Information found).
1601/01/01 09:30:00.0000000 872   1104                  Unknown( 39): GUID=50a5ef43-b2b6-3c52-145b-a386d8769665 (No Format Information found).

However, after investigating the registry keys on one of our servers, I saw a new registry key that I didn’t recognise, UpdateServiceUrlAlternate. See this dump from Powershell:

PS C:\Users\me\Desktop> ls HKLM:\SOFTWARE\Policies\Microsoft\Windows\


    Hive: HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows


Name                           Property
----                           --------
Appx
BITS                           EnableBitsMaxBandwidth     : 0
                               MaxBandwidthValidFrom      : 9
                               MaxBandwidthValidTo        : 17
                               MaxTransferRateOnSchedule  : 1000
                               MaxTransferRateOffSchedule : 9999
CurrentVersion
DataCollection
EnhancedStorageDevices         TCGSecurityActivationDisabled : 0
IPSec
Network Connections            NC_PersonalFirewallConfig : 0
NetworkConnectivityStatusIndic (default) :
ator
NetworkProvider
safer
SettingSync                    EnableBackupForWin8Apps : 1
System
WcmSvc
WindowsUpdate                  UpdateServiceUrlAlternate                    : http://localhost:8005
                               WUServer                                     : http://wsusserver.domainname.au:8530
                               WUStatusServer                               : http://wsus.domainname.au:8530
                               ElevateNonAdmins                             : 0
                               DoNotConnectToWindowsUpdateInternetLocations : 1
                               TargetGroupEnabled                           : 1
                               TargetGroup                                  : Auto Patch2
WorkplaceJoin                  (default) :
WSDAPI

After investigating this for a while, I found that the UpdateServiceUrlAlternate is a new setting introduced in the 2017 January update for Windows 10 1607 (and Server 2016, it would seem). For more details, see the Microsoft Documentation for UpdateServiceUrlAlternate.

We discovered that we had not been setting this with our current Group Policy, because we were using ADMX templates that were released before this was added. After updating our ADMX templates and setting that value to nothing, the updates download and install correctly again.

VMware User Environment Manager (UEM) Profile for Autodesk AutoCAD 2017 in VDI

In our Horizon View VDI environment we use UEM to manage some application’s user data. I used the Application Profiler included with UEM to create this profile for AutoCAD. One issue that we we have with AutoCAD is that it will reconfigure when launched by a new user. We tried to capture a default profile but the registry entries captured include lots of references to the specific user profile path for that user, which caused issues on launch. I have attached the INI file and icons to this post.

To prevent AutoCAD from crashing on launch, make sure to capture <LocalAppData>\Autodesk, which isn’t included by default when using the Application Profiler.

Application: Autodesk AutoCAD 2017

OS: Windows 7

INI Details:

# Flex config file generated with VMware User Environment Manager Application Profiler version 9.1

[IncludeRegistryTrees]
HKCU\Software\Autodesk
HKCU\SOFTWARE\FLEXlm License Manager

[IncludeFolderTrees]
<AppData>\Autodesk
# Added to prevent crash on launch
<LocalAppData>\Autodesk

[Metadata]
ProcessIfExists=
DirectFlexEnabled
DirectFlexPath=%ProgramFiles%\Autodesk\AutoCAD 2017\acad.exe
ProcessCriteria=
ProfilerChild=c:\program files\autodesk\autocad 2017\acwebbrowser\acwebbrowser.exe
ProfilerChild=c:\program files\autodesk\autocad 2017\admigrator.exe
ProfilerChild=c:\windows\microsoft.net\framework64\v4.0.30319\csc.exe
ProfilerChild=c:\windows\microsoft.net\framework64\v4.0.30319\cvtres.exe
ProfilerChild=c:\program files\autodesk\autocad 2017\dadispatcherservice.exe


[Immidio Flex]
This file was created using VMware UEM Management Console version 9.1.0.175.
Use only with VMware UEM.

Zipped INI File and Icon: Autodesk_AutoCAD_2017

Please feel free to comment with any issues or suggestions you had using AutoCAD with UEM in your environment.

Experiment: Using the Colemak Keyboard Layout For Two Weeks

Recently I have been investigating using better, more ergonomic computer equipment. To this end, several months ago I purchased an ergonomic keyboard, the Microsoft Natural Ergonomic Keyboard 4000. During my investigation of keyboards, however, I came across a lot of discussion about alternative keyboard layouts, and their various pros and cons, and have decided to try switching to the Colemak Keyboard layout.

About Me

I don’t have any issues with RSI, or pain or stiffness in my hands. I would like to keep it that way, and so I would like to use good ergonomic equipment and practices to keep my hands that way. I work in IT, and do a large amount of typing throughout the day, so even a small improvement to the quality of my typing can pay off for me.

Choosing Colemak

During my investigation on different keyboard layouts, I of course found a lot of information on Dvorak, another popular alternative layout. Certainly there are some significant benefits to Dvorak:

  1. It has more OS support (comes built-in on Windows)
  2. It’s more widely used, and therefore more studied.

So why choose Colemak?

  1. Most importantly for me, it keeps some of the keyboard shortcuts the same, like Ctrl-C, Ctrl-X and Ctrl-V. On Dvorak the common pattern Ctrl-C -> Ctrl-V are both two handed shortcuts. 🙁
  2. It has slightly less finger travel than Dvorak. For more details, see https://colemak.com/Ergonomic
  3. It keeps almost all the symbol keys in the same place, which is nice.

Day 1

The first day was so hard, and it went soooo slowly. Very frustrating, and by the end of the day I was getting only 12 WPM. I found I was constantly having to read off the image of the keyboard layout to type simple things.

Days 2 – 7

Once I got through the first day, I slowly improved over the next few days. I found that I was not having to look to remember the keys on the home row, and only check for keys above or below. However, even with the home row, I would get stuck with a mental block about once per sentence, trying to un-jumble the keys in my mind. This significantly slowed my typing, but as the week went on, the pauses would get shorter and less common.

Days 8 – 14

Once I was past the first week, I found I was no longer running into issues getting stuck, however my accuracy and speed were still lower than I had on QWERTY. However, as the week progresses I found my speed and accuracy picking up. I also started using the CapsLock backspace key, which is great.

Conclusion

So how fast am I typing now? I just tested my typing speed, and I’m getting 35 WPM. That’s not bad, but it’s not great. I have previously scored up to 50 WPM on QWERTY, although I haven’t tested my speed for a while. However, I plan to continue using Colemak, as I find it very comfortable to type on. We’ll see how I’m going after a couple of months of usage.

How to Opt-In and Detect to Microsoft Update for other Microsoft products using Powershell

Update 2018-03-25: Updated the script to make it more easily runnable from the command line (thanks Grant for the suggestion). I’ve also made the script available from on my Gitlab repo. To run the script from the command line, download it from the Gitlab repo, and then execute like this:

# CMD Prompt
powershell.exe -ExecutionPolicy Bypass -File C:\Configure-UpdateMSProducts.ps1 -RunType InstallService
# PowerShell Prompt
C:\Configure-UpdateMSProducts.ps1 -RunType InstallService

Update 2017-04-05: Thanks to some help from rog in the comments, I discovered there is a way to do this using group policy! 😀 To set this using group policy, go to Computer Configuration\Administrative Templates\Windows Components\Windows Update\ and configure the Configure Automatic Updates setting. In this setting there is a checkbox to set the Install updates for other Microsoft products setting. I will note, however, that this setting is not available in the local group policy editor for Windows 7, so I’m not sure if this setting would correctly apply to a Windows 7 machine. It’s definitely available on a Windows 10 1607 machine, though. Thanks for the help on this one, rog!

With Windows 10 out, we have been investigating how we can better deploy and manage Windows computers. Previously have deployed our images using an SCCM server, and then managed updates using WSUS integrated with SCCM, but Windows 10 brings a new update method, Windows Update for Business.

However, have you ever seen a checkbox with this label in your updates settings: Give me updates for other Microsoft products when I update Windows. How do you set this?

My first thought was to set this from group policy. Since this is how you set the Windows Update for Business settings, this would be the correct way to set the update method for other Microsoft products, right? So I have a look through the Windows Update, but can’t find the setting. I also do a little searching online, but all I can find is This Question on Microsoft’s forum. Here, JuliusPIV is asking the exact question I have.

So what’s the solution JuliusPIV found? This Microsoft page, Opt-In to Microsoft Update where Microsoft describes how to set enable it. Fixed, right?

There are two issues with this solution:

  1. It uses VBScript. I hate VB
  2. It doesn’t have any way to turn the setting back off again, or detect whether it is on or off

So I worked out how to do it in Powershell, and added a few extra options.

Solution:

Here’s how to turn it on

$ServiceManager = New-Object -ComObject "Microsoft.Update.ServiceManager"
$ServiceManager.ClientApplicationID = "My App"
$NewService = $ServiceManager.AddService2("7971f918-a847-4430-9279-4a52d1efe18d",7,"")

Here’s how to turn it off

$ServiceManager = New-Object -ComObject "Microsoft.Update.ServiceManager"
$ServiceManager.RemoveService("7971f918-a847-4430-9279-4a52d1efe18d")

Here’s a snippet of Powershell that will do whatever you want:

param(
    [String][ValidateSet('Detect','InstallService','RemoveService')]$RunType = "InstallService"
)

# Prepare a Windows Update service manager
$ServiceManager = New-Object -ComObject "Microsoft.Update.ServiceManager"

$ServiceID = "7971f918-a847-4430-9279-4a52d1efe18d"

# 
switch ($RunType)
{
    "Detect"
    {
        $found = $false
        foreach($service in $ServiceManager.Services)
        {
            if($service.ServiceID -eq $ServiceID)
            {
                $found = $true
            }
        }

        if($found)
        {
            "Found Service"
        }
    }
    "InstallService"
    {
        $ServiceManager.ClientApplicationID = "My App"
        try
        {
            $NewService = $ServiceManager.AddService2($ServiceID,7,"")
        }
        catch
        {
            Write-Warning "Failed to register service"
            Write-Warning $_.Exception.Message
            Exit 1
        }
        if($NewService.IsPendingRegistrationWithAU)
        {
            Write-Verbose "Needs to reboot"
            Exit 3010
        }
        else
        {
            Exit 0
        }
    }
    "RemoveService"
    {
        try
        {
            $ServiceManager.RemoveService($ServiceID)
        }
        catch
        {
            if($_.Exception.ErrorCode -eq -2145091564)
            {
                Write-Verbose "The service doesn't exist, so exit successfully"
                Exit 0
            }
            else
            {
                Write-Warning "Failed to remove service"
                Write-Warning $_.Exception.Message
                Exit 1
            }
        }
    }
    default
    {
        Write-Warning "No RunType set. Exiting"
        Exit 1
    }
}