Fixing WSL Ubuntu on Windows 11 Insider Preview

2024-11-24T00:00:00+00:00

I've had an issue for a month or so where WSL hasn't been working right, but I've been working around it by getting more comfortable with PowerShell, doing a substantial amount of stuff over SSH to a DigitalOcean Droplet, doing work on other (Linux) computers in the house, and as a last resort... doing development work in raw Windows. It's a bit less painful that I thought it would be, especially for languages like Python, Go, and Rust.</p>

What I've missed the most is the ability to run k8s with Rancher Desktop and the ability to try out new things like Podman. I'm finally getting around to fixing it, and what follows is basically a summary of what I learned, and then a stream of consciousness of what I did to fix it.</p>

TLDR</h2>

TLDR of things I learned while looking into this issue:</p>

Hyper-V

Windows / Hyper-V Firmware Signatures typically expire on 9/15</li>
You can workaround an outdated Hyper-V firmware signature by manually setting the datetime to something before it expired</li> </ul> </li>

Windows Update

This tool can lie to you and get stuck in states where the main fix is to turn off all the relevant services, clear their caches, and turn the services back on.</li> </ul> </li>

Windows Error Codes

Some of these have substantial first-party documentation with detailed troubleshooting steps and explanations</li>
Some have no first-party documentation, but sometimes a kind internet user with "-MSFT" at the end of their username on a Microsoft forum will say "do these actions to fix it" and not explain why.</li> </ul> </li>

Windows Performance Analyzer

This tool exists, and can be used to read ETL files.</li>
It's a bit tricky to read, but it can give you insight into which Windows source files are causing a problem, but you still can't read the code to figure it out for yourself.</li> </ul> </li>

microsoft/WSL on Github

Provides a nice library of diagnostic/log collection PowerShell scripts you can use to dig deeper into things.</li>

Sometimes people have reported and had the same issue as you already fixed so you can just borrow the solution from there.</li> </ul> </li> </ul>

Initial Diagnostic Efforts</h2>

When attempting to run WSL Ubuntu I get the following message</p>

Invalid Signature.</span></span>
Error code: Wsl</span>/</span>Service</span>/</span>CreateInstance</span>/</span>CreateVm</span>/</span>HCS</span>/</span>0x80090006</span></span>
</span>
[</span>process exited with code</span> 4294967295</span> (</span>0xffffffff</span>)]</span></span>
You can now close this terminal with Ctrl</span>+</span>D</span>,</span> or press Enter to restart.</span></span></code></pre>
Some Diagnostic Efforts:</p>
(base) PS C:\Users\jayml</span>></span> wsl </span>--</span>help</span></span>
Output Omitted </span>for</span> Brevity</span></span>
</span>
(base) PS C:\Users\jayml</span>></span> wsl </span>--</span>status</span></span>
Default</span> Distribution: Ubuntu</span></span>
Default</span> Version: </span>2</span></span>
</span>
(base) PS C:\Users\jayml</span>></span> wsl </span>--</span>version</span></span>
WSL version: </span>2.3</span>.</span>26.0</span></span>
Kernel version: </span>5.15</span>.</span>167.4</span>-</span>1</span></span>
WSLg version: </span>1.0</span>.</span>65</span></span>
MSRDC version: </span>1.2</span>.</span>5620</span></span>
Direct3D version: </span>1.611</span>.</span>1</span>-</span>81528511</span></span>
DXCore version: </span>10.0</span>.</span>26100.1</span>-</span>240331</span>-</span>1435.</span>ge</span>-</span>release</span></span>
Windows version: </span>10.0</span>.</span>26217.5000</span></span></code></pre>winver</span></span></code></pre>
</p>
Collecting & Peeping WSL Logs</h2>
Not seeing anything super useful or how to get logs from the help I found a guide from Microsoft Troubleshooting Windows Subsystem for Linux</a> and two existing Github Issues on the microsoft/WSL project. I did skim the issues to see what was going on, but I didn't thoroughly read them at first because I want to see how far I can get in working it out myself. (with help from documentation :D)</p>

Ubuntu stopped working with code 0x80090006</a></li>
WSL suddenly stopped working error code - Error: 0x80090006 Invalid Signature.</a></li>
</ul>
Microsoft has a bot that explains how to get the logs by running a remote Powershell script as admin</p>
</p>
Copy & Paste-able text here for ease of use if you need it:</p>
Invoke-WebRequest</span> -</span>UseBasicParsing </span>"https://raw.githubusercontent.com/microsoft/WSL/master/diagnostics/collect-wsl-logs.ps1"</span> -</span>OutFile collect</span>-</span>wsl</span>-</span>logs.ps1</span></span>
Set-ExecutionPolicy</span> Bypass </span>-</span>Scope </span>Process</span> -</span>Force</span></span>
.\collect</span>-</span>wsl</span>-</span>logs.ps1</span></span></code></pre>
And the output from actually running it:</p>
Log collection is running. Please reproduce the problem and once done press any key to save the logs.</span></span>
Saving logs...</span></span>
100</span>%</span>  [</span>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>></span>]</span></span>
Logs saved </span>in</span>: C:\Users\jayml\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02.</span>zip. Please attach that file to the GitHub issue.</span></span></code></pre>
That lead me to a nice collection of diagnostic scripts in microsoft/WSL/blob/master/diagnostics</a> which could be useful in the future :)</p>
I want to poke around in the log file and see if I can figure out what's going on before I finish reading the GitHub issues.</p>
After unzipping it I get the following directory:</p>
(base) PS C:\Users\jayml\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span>\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span>></span> ls</span></span>
</span>
    Directory: C:\Users\jayml\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span>\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span></span>
</span>
Mode                 LastWriteTime         Length Name</span></span>
----                 -------------         ------ ----</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM            </span>650</span> acl.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>1670</span> appxpackage.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>2562</span> bcdedit.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>1356</span> HKCU.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>1582</span> HKLM.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM        </span>3670016</span> logs.etl</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM          </span>16930</span> optional</span>-</span>components.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>3038</span> P9NP.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM       </span>30353928</span> windows</span>-</span>version.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM         </span>210532</span> Winsock2.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM            </span>160</span> wpr.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM           </span>9342</span> wsl.wprp</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM            </span>990</span> wslservice.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM            </span>360</span> wslsupport</span>-</span>impl.txt</span></span>
-----</span>          11</span>/</span>24</span>/</span>2024  4</span>:</span>52</span> PM            </span>574</span> wslsupport</span>-</span>proxy.txt</span></span></code></pre>
Not wanting to look through all the files individually I tried "grepping" for a handful of different things until I found the following</p>
(base) PS C:\Users\jayml\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span>\WslLogs</span>-</span>2024</span>-</span>11</span>-</span>24_16</span>-</span>52</span>-</span>02</span>></span> Get-ChildItem</span> -</span>Path </span>"."</span> |</span> Select-String</span> -</span>Pattern </span>"CreateVm"</span></span>
</span>
logs.etl:</span>381</span>:�</span></span>
@@</span></span>
44Microsoft.Windows.Lxss.Manager␦sPOω�G�����v�</span>8</span></span>
))CreateVmBeginwslVersionPart</span></span>
A_PrivTags</span></span></code></pre>
It's clear there's some non-printable/binary sequences going on in that .etl</code> file, and I only know ETL as Extract-Transform-Load, so I wasn't really sure what this was or how to open it.</p>
Windows Performance Analyzer?</h2>
Once again Microsoft docs pull through with Opening and Analyzing ETL Files in WPA</a> a tool that I had never heard of until now.</p>
After opening the ETL file in WPA (Windows Performance Analyzer) it looks like this</p>
</p>
So I need to click around a bit and get my bearings.</p>
I had to double click on the "System Activity" section of "Graph Explorer - logs.etl" to get it to start showing more useful looking stuff.</p>
</p>
There is a search function that lets me search for CreateVm</p>
</p>
I can see some source files and line numbers where things might be going wrong, but I have no way of actually inspecting that code to see what's happening.</p>
</p>
There even seems to be an error while attempting to find the error message</p>
winrt::hresult_error: The text associated with this error code could not be found.</span></span></code></pre>
I was able to poke through and find ever-so-slightly more specific details like the following:</p>
onecore\vm\dv\chipset\bios\biosdevice.cpp(411)\vmchipset.dll!00007FFE0C4D91FF: (caller: 00007FF66009D42E) ReturnHr(1) tid(18f8) 80090006 Invalid Signature.  </span></span>
   Msg:[onecore\vm\common\guestloader\lib\vmfirmwareloader.cpp(79)\vmchipset.dll!00007FFE0C44CA46: (caller: 00007FFE0C47FA9F) Exception(1) tid(18f8) 80090006 Invalid Signature.  </span></span>
   Msg:[signature state 2]  </span></span>
]</span></span></code></pre>
</p>
Revisiting GitHub Issues</h2>
I just don't seem to be able to trace it to an actual cause, so basically I'm stuck. Although I got a much deeper peek into how Windows is working under the hood than I ever remember being possible. At this point it's back to review the issues filed on Github.</p>
Someone (Microsoft Employee?) suggests that cmfirmwareloader.cpp:79 "signature state 2"</code>  means that the Hyper-V Firmware is expired.</p>
</p>
And someone else suggested that manually setting the time to a date prior to what shows up in the winver</code> command will allow it to work.</p>
</p>
Which... appears to be true when I tried it.</p>
</p>
Although I do have a Canary update available that I'm going to try and install and see if that works!</p>
Windows 11 Insider Preview 27729.1000 (rs_prerelease)</span></span></code></pre>
After rebooting it seems the update didn't take.... If I check my update history, there's actually several failed attempts to install this update</p>
</p>
Fixing Windows Update</h2>
Looking up the code 0xc1900101</code> I found the following two sources from Microsoft</p>

Get help with Windows upgrade and installation errors</a></li>
Windows 10 upgrade resolution procedures</a></li>
</ul>
One of them said to check these some logs in C:\$WINDOWS.~BT\Sources\Rollback</code> for an extended error code, to get more specific with it, but the exact file wasn't actually located in that directory so I couldn't do that.</p>
The Checklist</h3>
Both seem to indicate that it's likely an issue with an incompatible driver. And recommend the following steps.</p>


Have 20GB Free for Updating a 64-bit OS</li>

Run Windows Update a few times</li>

Check third-party drivers and download any updates</li>

Unplug Extra Hardware</li>

Check Device Manager for Errors</li>

Remove third-party Security Software</li>

Repair Hard Drive Errors chkdsk/f C:</code> (from cmd.exe)</li>

Do a Clean Restart into Windows</li>

Restore and Repair System Files dism /Online /Cleanup-Image /Restorehealth</code></li>
</ul>
(I'm returning back to this checklist and checking off items as I do them, in all it took me around an hour to go through everything. Clean restart turned out to be unnecessary in this case.)</p>
I've used DISM before, and I know it can be running for quite a while, so I'm going to kick it off in the background while I start doing the other checklist items.</p>
Updating Drivers & Checking for Errors in Device Manager</h3>
Since I can use device manager for Checking and Updating Drivers, as well as looking for errors (and it keeps me from bending over to unplug things for a few moments longer) that's where I'll start.</p>
Weirdly when I tried to launch it both from the "Start Menu", and Win+R Run as devmgmt.msc</code> I get blocked by UAC claiming "An Administrator has blocked you from running this app" which, since I am the administrator is... highly suspicious</em>, especially since I checked the settings by viewing the Properties of that file in System32.</p>
TODO: Insert photo taken from phone because I can't screenshot UAC prompts in any meaningful way.</p>
I was able to launch it successfully from an Admin PowerShell and I don't see any errors or updates right away, but one of the first things I notice is that I have VoiceMeeter installed and I haven't used that in a few years so I figure I can go ahead and remove it. Both through the Uninstall Apps feature, and Driver-by-Driver if I have to. (I didn't have to, using the uninstaller cleared up all the drivers.)</p>
</p>
Also noticed there are some Bluetooth drivers for devices that I still have but haven't used with this system in a number of years, so I'm removing those as well. From things like Aftershokz or Beats headphones that I use in other contexts. They showed up under "Bluetooth" and "Sound, video and game controllers" and "System devices" with slightly different, but recognizably similar names.</p>
After clearing all that out and rescanning, still no errors or updates, and DISM completed in the background. Now I'm checking for any 3rd-party security software before I start unplugging anything.</p>
Uninstalling Third-Party Security Software</h3>
winget list</code> is a convenient way to list all the installed packages even if they aren't managed by winget. Running that command and skimming the list I found the following that could be described as "third-party security software":</p>
- Bitwarden 2024.10.1</span></span>
- Mullvad VPN 2023.3.0</span></span>
- OWASP Zed Attack Proxy 2.13.0</span></span>
- RustDesk 1.1</span></span>
- qFlipper 1.3.3</span></span>
- KeePassXC 2.7.7</span></span>
- Malwarebytes version 4.6.17.334</span></span>
- Tailscale 1.76.6</span></span>
- OpenSC smartcard framework (64bit) 0.22.0.0</span></span>
- sniffnet 1.2.2</span></span>
- TightVNC 2.8.59.0</span></span>
- Angry IP Scanner 3.8.2</span></span>
- Mudfish Cloud VPN v5.7.6</span></span>
- Npcap 1.78</span></span>
- OSSEC HIDS 3.7.0</span></span>
- Portmaster 0.8.8.1</span></span>
- TeamViewer 15.51.6</span></span>
- Wireshark 4.2.5 x64</span></span>
- YubiKey Manager 1.2.4</span></span>
- Tactical RMM Agent 1.8.0</span></span>
- STIG Viewer 3 (Machine) 3.3.0</span></span>
- BleachBit 4.4.2.2142 (current user) 4.4.2.2142</span></span>
- mitmproxy 11.0.0.0</span></span></code></pre>
I also noticed some other potentially relevant software (as-in "could easily have outdated drivers")</p>
- X56 H.O.T.A.S 8.0.213.0</span></span>
- ASIO4ALL 2.15</span></span>
- Peace 1.6.4.1</span></span>
- Equalizer APO 1.2.1</span></span>
- OnePlus USB Drivers 1.00</span></span>
- Samsung Data Migration 4.0.0.18</span></span>
- Qualcomm USB Drivers For Windows 1.00.57</span></span></code></pre>
Some of these tools I'm actively using, some of these tools I haven't used in a long time, and some of these tools I'm happy to forget about or install again if it turns out I need them. So I'll start uninstalling and record what I've removed here.</p>
- X56 H.O.T.A.S.</span></span>
- ASIO4ALL, Peace, EqualizerAPO (All Audio Software)</span></span>
- OnePlus & Qualcomm USB Drivers</span></span>
- Mudfish Cloud VPN (removed a TAP-Win32 Adapter V9 driver while uninstalling)</span></span>
- RustDesk, TeamViewer, TightVNC, Tactical RMM Agent</span></span>
- OSSEC HIDS, Sniffnet, OWASP ZAP, mitmproxy</span></span>
- Malwarebytes</span></span></code></pre>
After removing those I ran another winget list</code> to skim through as a check, and it looks good to me, I know there are still software installed with third party drivers e.g. Wireshark+Npcap, but I'm unwilling to uninstall them unless it's actually as a last resort.</p>
Repairing Hard Drive Errors</h3>
At this point I decided to drop into a cmd.exe</code> from the Admin PowerShell to repair hard drive errors, since I figured that was something I could do before unplugging things as well... and I discovered that I can't check the disk while another process is using it, which means this can only occur during reboot.</p>
(base) PS C:\Users\jayml</span>></span> cmd</span></span>
Microsoft Windows [</span>Version</span> 10.0</span>.</span>26217.5000</span>]</span></span>
(c) Microsoft Corporation. All rights reserved.</span></span>
</span>
C:\Users\jayml</span>></span>chkdsk</span>/</span>f C:</span></span>
The type of the file system is NTFS.</span></span>
Cannot lock current drive.</span></span>
</span>
Chkdsk cannot run because the volume is </span>in</span> use by another</span></span>
process</span>.  Would you like to schedule this volume to be</span></span>
checked the next time the system restarts? (Y</span>/</span>N) Y</span></span>
</span>
This volume will be checked the next time the system restarts.</span></span></code></pre>Finally unplugging devices</h3>
May as well unplug some devices now. When I checked, all that I have plugged in that's accessible from outside the chassis was the following:</p>

Keyboard</li>
Mouse</li>
Microphone</li>
YubiKey</li>
2x Display Port Monitors
I keep my webcam unplugged because I use it infrequently.</li>
</ul>
So I unplugged the Microphone and YubiKey, since I need the Mouse & Keyboard to actually do anything else.</p>
Gonna "Restart Now" from Windows Update and see if the update or chkdsk</code> takes.. And it turns out the chkdsk did start scanning the drive while rebooting. So far I’ve seen Stage 1 but I missed the name and can’t find the names of the stages online and “Fixing (C:) Stage 2” I’m not actually certain whether that implies any errors or if it always says that.</p>
Still doesn't work, but yay new error code!</h3>
This time it did finish booting rather than hanging at the splash screen. But, there was still an error when attempting to install the update, this time a new error code 0x80248007</code> AKA "We made progress!"</p>
The same page listing details for the 0xc1900101</code> error code also lists some 0x8024</code> error codes but not the exact one I'm seeing. When searching I can't find a completely clear first-party source that explains what this error code 0x80248007</code> means and exactly how to fix it. However I did find on the Microsoft Learn Forum</a> someone asked a question about it, and user AllenLiu-MSFT left an answer which I cannot figure out how to permalink directly to, but it contains the following advice.</p>

Run the Windows Update Troubleshooter</li>
Clear the Windows Update Cache</li>
Reset the Windows Update Components</li>
Perform a clean boot and then try to install the update</li>
Manually download and install the update from the Microsoft Update Catalog</li>
</ol>
For running the Windows Update troubleshooter their recommended path is
"Settings > Update & Security > Troubleshoot > Additional troubleshooters > Windows Update > Run the troubleshooter"
...following that breadcrumb trail fails for me instantly as the path I take is
"Settings > Windows Update > ???"
There's a "Get Help" button which takes me to a blank white screen...
But under Windows Update if I go to "> Advanced Options > Recovery" I have a "Fix problems without resetting your PC" option, which seems to be a troubleshooter of some kind. Although I can now see it has taken me to a totally different place called "Settings > System > Recovery" and this gives me an "Other troubleshooters" option that I can use to launch the "Windows Update Troubleshooter" which actually does appear to take me to the "Get help" screen I originally saw, except now it's a search box, and it's asking me to sign in.</p>
</p>
I am able to sign in and run it, but the actual tool itself failed to run.</p>
</p>
I don't really want to hand off to an agent (even though I know I won't change my mind about this I'm leaving the window open in the background just in case) so...</p>
Clearing the Windows Update Cache (This actually fixed things!)</h3>
Moving down the list to "Clear the Windows Update Cache", which AllenLiu-MSFT recommends running the following from an Administrator Command Prompt.</p>
net</span> stop wuauserv</span></span>
net</span> stop cryptSvc</span></span>
net</span> stop bits</span></span>
net</span> stop msiserver</span></span>
ren</span> C:</span>\W</span>indows</span>\S</span>oftwareDistribution SoftwareDistribution.old</span></span>
ren</span> C:</span>\W</span>indows</span>\S</span>ystem32</span>\c</span>atroot2 catroot2.old</span></span>
net</span> start wuauserv</span></span>
net</span> start cryptSvc</span></span>
net</span> start bits</span></span>
net</span> start msiserver</span></span></code></pre>
This "removes" those two files from where the Windows Update service looks for them by renaming them, this is presumably so we could restore them if desired. It also stops and starts the following:</p>

Windows Update service</li>
Cryptographic Services service</li>
Background Intelligent Transfer Service service</li>
Windows Installer service</li>
</ul>
It might not mean anything but after stopping wuauserv</code> and cryptSvc</code> both bits</code> and msiserver</code> reported that they were not started. When I went to start everything again they seem to have started just fine. It's possible those service processes die when they don't have one or both of the Update or Cryptographic service running.</p>
Since we ran those commands I'm going to hit the Retry button in Windows Update and see how it goes.</p>
It returned a ton of new update packages that weren't available before including a slightly newer preview build Windows 11 Insider Preview 27754.1000 (rs_prerelease)</code> I have high hopes for the install succeeding this time, but I'm willing to keep moving through AllenLiu-MSFT's list until it works :D</p>
</p>
Most of those downloaded and installed successfully and now I get to restart again and see if the Windows 11 Update works this time..</p>
</p>
It's Alive!</h3>
After rebooting it does actually seem to be applying updates this time, that’s another new thing! And once fully rebooted wSinver shows that the Eval copy expires 9/15/2025</code> and WSL/Ubuntu properly loads while automatically set to the current datetime.</p>
</p>
After all this there does appear to still be a DNS resolution issue in WSL/Ubuntu, but I don't actually need DNS resolution for what I wanted this for, so that will be for future me to figure out when I need it.</p>
Also, happy to report that I'm now able to run devmgmt.msc</code> through both the Start Menu and Win+R without an unwelcome visit from the UAC fairies.</p>
It's unrelated but a little bit curious that during this investigation my WiFi Router/Switch reported that the internet was disconnected, but my connection appeared to be working fine.</p>
</p>

Testing Python Code

2019-05-28T00:00:00+00:00

The following is a selection of some Python testing libraries, commandline tools, etc. that I use, find cool, and have found real utility in when testing Python Code.</p>

unittest</h3>

The unittest module</a> ships with python since 2.7 and includes some basic test discovery</a>.</p>

import</span> unittest</span></span>
</span>
class</span> TestStringMethods</span>(</span>unittest</span>.</span>TestCase</span>):</span></span>
</span>
    def</span> test_upper</span>(</span>self</span>):</span></span>
        self</span>.</span>assertEqual</span>(</span>'foo'</span>.</span>upper</span>(),</span> 'FOO'</span>)</span></span>
</span>
    def</span> test_isupper</span>(</span>self</span>):</span></span>
        self</span>.</span>assertTrue</span>(</span>'FOO'</span>.</span>isupper</span>())</span></span>
        self</span>.</span>assertFalse</span>(</span>'Foo'</span>.</span>isupper</span>())</span></span>
</span>
if</span> __name__</span> ==</span> '__main__'</span>:</span></span>
    unittest.</span>main</span>()</span></span></code></pre>doctest</h3>
The doctest module</a>
also comes with python, but works in an entirely different way.
It reads docstrings and parses them for content that looks like an
interactive python session. It will try to execute the line that looks
like someone typed it, and compare it to the line that looks like the
expected result.</p>
def</span> factorial</span>(</span>n</span>):</span></span>
    """Return the factorial of n, an exact integer >= 0.</span></span>
</span>
    >>> </span>[factorial(n) for n in range(6)]</span></span>
    [1, 1, 2, 6, 24, 120]</span></span>
    >>> </span>factorial(30)</span></span>
    265252859812191058636308480000000</span></span>
    >>> </span>factorial(-1)</span></span>
    Traceback (most recent call last):</span></span>
        ...</span></span>
    ValueError: n must be >= 0</span></span>
</span>
    Factorials of floats are OK, but the float must be an exact integer:</span></span>
    >>> </span>factorial(30.1)</span></span>
    Traceback (most recent call last):</span></span>
        ...</span></span>
    ValueError: n must be exact integer</span></span>
    >>> </span>factorial(30.0)</span></span>
    265252859812191058636308480000000</span></span>
</span>
    It must also not be ridiculously large:</span></span>
    >>> </span>factorial(1e100)</span></span>
    Traceback (most recent call last):</span></span>
        ...</span></span>
    OverflowError: n too large</span></span>
    """</span></span>
</span>
    import</span> math</span></span>
    if not</span> n</span> >=</span> 0</span>:</span></span>
        raise</span> ValueError</span>(</span>"n must be >= 0"</span>)</span></span>
    if</span> math.</span>floor</span>(n)</span> !=</span> n:</span></span>
        raise</span> ValueError</span>(</span>"n must be exact integer"</span>)</span></span>
    if</span> n</span>+</span>1</span> ==</span> n:</span>  # catch a value like 1e300</span></span>
        raise</span> OverflowError</span>(</span>"n too large"</span>)</span></span>
    result</span> =</span> 1</span></span>
    factor</span> =</span> 2</span></span>
    while</span> factor</span> <=</span> n:</span></span>
        result</span> *=</span> factor</span></span>
        factor</span> +=</span> 1</span></span>
    return</span> result</span></span>
</span>
</span>
if</span> __name__</span> ==</span> "__main__"</span>:</span></span>
    import</span> doctest</span></span>
    doctest.</span>testmod</span>()</span></span></code></pre>
I think this is useful as expressive documentation of
some use cases of a module or function, but not as a primary unit-testing
strategy. Essentially this just confirms that the examples in your
documentation do work, and proof of that can actually be really strong if
used well.</p>
py.test</h3>
Pytest</a> is the poster-child awesome
testing project that is just more convenient to use than unittest, and has many
plugins available to support any kind of scenario you could come up with. A
huge convenience comes in the form of how it actually renders the output in your console.</p>
# content of test_sample.py</span></span>
def</span> inc</span>(</span>x</span>):</span></span>
    return</span> x</span> +</span> 1</span></span>
</span>
</span>
def</span> test_answer</span>():</span></span>
    assert</span> inc</span>(</span>3</span>)</span> ==</span> 5</span></span></code></pre>
From this example you can see it's very simple to get started with pytest, even more simple than unittest.
Additionally, you can create some very complicated and expressive testing infrastructure with pytest.
In fact literally every other testing module mentioned here can be integrated with pytest in a convenient way.</p>
hypothesis</h3>
The hypothesis</a> module is almost like a fuzzer-gone-testing-framework.
It helps to generate lots of different input meeting some defined constraints and essentially enables you to fuzz your
program with each test run. It only reports on failing inputs, and will almost certainly help you catch edge-cases you
wouldn't otherwise catch.</p>
The following code will generate input that is considered a valid str in python.</p>
from</span> hypothesis</span> import</span> given</span></span>
from</span> hypothesis.strategies</span> import</span> text</span></span>
</span>
@given</span>(</span>text</span>())</span></span>
def</span> test_decode_inverts_encode</span>(</span>s</span>):</span></span>
    assert</span> decode</span>(</span>encode</span>(s))</span> ==</span> s</span></span></code></pre>
I have friends who will argue that testing at least from frameworks should be idempotent or a functional map of a state of some code.
Mostly I think that's a cop-out and it seems silly. Hypothesis returns enough information for you to reproduce the errors that it created.
And it may create some errors that a human programmer isn't going to define on their own and overall I think it's a net benefit.</p>
tox</h3>
Tox</a> is a sort of automation tool I think originally created for CI pipelines which allows you to
test your python code against multiple versions of python with ease. It uses multiple actual versions of python and also
virtual environments to isolate dependencies for each instance of a test run. This can help make sure your code is compatible
with Python 2 and Python 3. However, Python 2 is deprecated officially January 1, 2020 so ideally projects will no longer
need this kind of compatibility testing, although I'm sure there will be at least a handful of projects that remain
supported in that way for a while.</p>
That being the state of things, not everyone needs this, but I like using it anyway.</p>
To choose which versions of python you want to test against, you create a file called tox.ini in the same dir as your
setup.py</code> file.</p>
[tox]</span></span>
envlist</span> =</span> py27,py36</span></span>
</span>
[testenv]</span></span>
# install pytest in the virtualenv where commands will be executed</span></span>
deps</span> =</span> pytest</span></span>
commands</span> =</span></span>
    # NOTE: you can run any command line tool here - not just tests</span></span>
    pytest</span></span></code></pre>Coverage.py</h3>
Coverage.py</a> is a code coverage tool,
it measures code coverage with runtime analysis. It produces reports that look like this</p>
$</span> coverage report</span> -m</span></span>
Name</span>                      Stmts   Miss  Cover   Missing</span></span>
-------------------------------------------------------</span></span>
my_program.py</span>                20      4</span>    80%   33-35,</span> 39</span></span>
my_other_module.py</span>           56      6</span>    89%   17-23</span></span>
-------------------------------------------------------</span></span>
TOTAL</span>                        76     10</span>    87%</span></span>
</span></code></pre>
This can useful for making guarantees that your test suite is exercising enough of the code to give you a necessary level of confidence.
I know a lot of people chase code coverage by attempting to make a 1 to 1 mapping of tests and units, but I think
that just leads to low quality tests. One of the reasons I like hypothesis</code>, and also the strategy, Behavior Driven Development
is that it makes more sense to test that a "larger" portion of your code is exhibiting the correct and consistent Behavior
than it is to ensure that each small piece of code has a test case related to it in some way.</p>
Radon</h3>
Radon</a> is a Comprehensive metrics suite giving you Cyclomatic Complexity,
Maintainability Index, and other things about your code that can be statistically analyzed.</p>
The accuracy and utility of the formulae used to produce many of these metrics may be questionable but I still find it valuable
as a general guide to helping me produce clearer and more concise code.</p>
By default the command line tool gives letter grades to your modules and functions/methods, which looks like this</p>
$</span> radon cc sympy/solvers/solvers.py</span> -a -nc</span></span>
sympy/solvers/solvers.py</span></span>
    F</span> 346:0 solve - F</span></span>
    F</span> 1093:0 _solve - F</span></span>
    F</span> 1434:0 _solve_system - F</span></span>
    F</span> 2647:0 unrad - F</span></span>
    F</span> 110:0 checksol - F</span></span>
    F</span> 2238:0 _tsolve - F</span></span>
    F</span> 2482:0 _invert - F</span></span>
    F</span> 1862:0 solve_linear_system - E</span></span>
    F</span> 1781:0 minsolve_linear_system - D</span></span>
    F</span> 1636:0 solve_linear - D</span></span>
    F</span> 2382:0 nsolve - C</span></span>
</span>
11</span> blocks (classes, functions, methods) analyzed.</span></span>
Average</span> complexity: F (61.0)</span></span></code></pre>
It can be coerced into giving you more information or qualitatively different kinds of information to better guide the
changes that would improve those scores. Usually a low score is due to something like a deeply nested loop or conditional.</p>

Jay Looney

Fixing WSL Ubuntu on Windows 11 Insider Preview

Find Your Element by Ken Robinson

Google BeyondCorp Notes - Paper 1

Purple Cow by Seth Godin

Start With Why by Simon Sinek

Mechanics of an Assembly Program

Testing Python Code

Jay Looney

Fixing WSL Ubuntu on Windows 11 Insider Preview

Fixing Windows Update</h2> Looking up the code 0xc1900101</code> I found the following two sources from Microsoft</p>

Find Your Element by Ken Robinson

Google BeyondCorp Notes - Paper 1

Purple Cow by Seth Godin

Start With Why by Simon Sinek

Mechanics of an Assembly Program

Testing Python Code