ads' corner

A while ago we bought new Pulse 15 laptops from Tuxedo Computers.

Installed Ubuntu on them, and for a while the sleep mode was nagging me, it does not really work. Today I found time to investigate this issue.

First let's find out which sleep mode is enabled:

cat /sys/power/mem_sleep
[s2idle]

Something is wrong here. Not only can't I set the sleep mode to the desired "deep" mode, the "deep" mode is not even available. That explains why things are not working as expected.

After some research it turns out that this might be a BIOS problem. Prepare to reboot!

After the laptop powers up, press the "Del" key until the BIOS comes up. Then navigate to "Advanced Settings" -> "AMD PBS" -> "S3/Modern Standby Support".

The default is "Modern Standby Enable", but it turns out that this is the culprit. Changing this to "S3 Enable" and then reboot fixes the problem and "deep" now shows up in the available Linux Power States.

Then follow the description in this link, and permamently set the sleep mode to "deep".

cat /sys/power/mem_sleep
s2idle [deep]

On my Mac, one of the annoying "features" is when the Mac screensaver comes on, the device eventually goes to sleep, and it disconnects the network. Which in turn timeouts services like Slack or Google, because these services keep a network connection open at all times. When waking up the device, I often have to login again into all the services, even though the device is just sitting in my working room on the desk all day and night. Very annoying.
I suppose it's one of these things where Apple thinks they know better how users want their device to behave.

KeepingYouAwake is a nice little tool which prevents all of this.

When it is running and activated, it prevents the Mac from going to sleep. Which in turn never deactivates the network. And never timeouts the online services.

What's not to love about it?

Picture made by Anton Atanasov

With people moving away from Twitter, mostly to Mastodon, discovering the new accounts became a problem.

For people in the PostgreSQL community I created a website which lists different social media accounts. This website is part of the "PostgreSQL Person of the Week" interview project, however the data source is dynamic, and stored in a different repository. This allows me to keep the repository for the website private, but publish the data for the social media links - this data is public anyway. The interview repository is private, because who wants to see upcoming interviews anyway? ;-)

The interview website is made with Hugo, a static website generator. Normally Hugo looks for content, templates, and other data in the current directory - my private repository.

As part of compiling the website, Hugo can fetch external data, either in JSON or CSV format. This is using the getJSON() and the getCSV() functions, which can be used in Shortcodes, as example.

If you use GitHub Actions to run Workflows and tests, you might have spotted this warning recently:

Node.js 12 actions are deprecated. For more information see: https://github.blog/changelog/2022-09-22-github-actions-all-actions-will-begin-running-on-node16-instead-of-node12/. Please update the following actions to use Node.js 16: actions/checkout@v2

This warning means that GitHub will deprecate a certain action, which checks out the repository into the runner. This is going on since early 2022 and by summer 2023 they plan to upgrade all actions to v16.

If you use GitHub Actions to run Workflows and tests, you might have spotted this warning recently:

The `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/

This warning means that GitHub will deprecate a certain syntax which populates variables, and disables it by end of May 2023.

For the "PostgreSQL Person of the Week" interviews I'm using Hugo as static blogging engine. Part of every interview is the list of tags, which links this interview to other similar interviews. However until recently no one really knew if a tag is popular or just used in this interview. I wanted to change this, and add the tag count behind every tag.

The way I use Hugo I have the previews also online, and I don't want to count any interviews which are not published (still drafted) when counting entries for a tag.

Tags in Hugo are taxonomies, which is a user-defined grouping of content.

While writing and stress-testing a Python script which uses the multiprocessing library, I ran into the problem that occasionally the script hangs at the end. Literally runs past the last line of code and then hangs.

In this script I'm using Queues and Events, so I made sure that I properly close the queues in the forked/spawned (tried both) processes and also clean out all queues in the parent. Nevertheless occasionally - like seldom, but it happens - the script hangs. Checked the process list, checked remaining threads, checked the queues, all fine. Still ...

When I hit Ctrl+C, I get the following stack trace:

^CError in atexit._run_exitfuncs:
Traceback (most recent call last):
  File "/usr/lib/python3.9/multiprocessing/util.py", line 300, in _run_finalizers
    finalizer()
  File "/usr/lib/python3.9/multiprocessing/util.py", line 224, in __call__
    res = self._callback(*self._args, **self._kwargs)
  File "/usr/lib/python3.9/multiprocessing/queues.py", line 201, in _finalize_join
    thread.join()
  File "/usr/lib/python3.9/threading.py", line 1053, in join
    self._wait_for_tstate_lock()
  File "/usr/lib/python3.9/threading.py", line 1073, in _wait_for_tstate_lock
    if lock.acquire(block, timeout):
KeyboardInterrupt

(Can't change the Python 3.9, as I don't control the target system where this is supposed to run)

There's nothing left to join, the only two remaining threads are MainThread and QueueFeederThread.

After some more debugging I had the simple idea not only to empty the queue before finishing the script, but setting the queue variable to None:

worker_queue = None

This helps. The script is no longer hanging at the end. At a later point I will see if the problem is fixed in newer versions.

When I make changes in OpenStreetMap, I often need to extract the object colour from a picture or a video, as example for a building, bench or a roof. This collides with a feature in KDE which I really like: Dim Inactive Windows.

The setting is in the KDE settings, under Desktop Effects -> Dim Inactive. As the name implies, it dims (makes them a bit more dark) all non-active windows - which makes it visually clear which window is currently active. Quite useful.

For picking a colour from a picture/video I'm using KColorChooser, which provides the colour in Hex code - exactly what I need in OSM.

However when KColorChooser is the active window, and all other windows are dimmed, it extracts the wrong colour from the picture/video - the colour which is already dimmed. Not what I need. Therefore when I do OSM edits, I temporarily disable this feature.

At first I went into the KDE settings, twice, but this is too many clicks. Then I figured out the commandline arguments to disable and enable this feature.

Disable "Dim Inactive":

kwriteconfig5 --file kwinrc --group Plugins --key diminactiveEnabled false
qdbus org.kde.KWin /KWin reconfigure

Enable "Dim Inactive":

kwriteconfig5 --file kwinrc --group Plugins --key diminactiveEnabled false
qdbus org.kde.KWin /KWin reconfigure

Now all I need to do is execute a small script which is in my personal ~/bin directory. Let the edits continue.

Note: I wish JOSM would have an integrated video player which also allows to pick colours. But so far I haven't found any such plugin.

Shortcodes in Hugo are a neat and poweful system to avoid repating the same piece of text over and over again. Let's say I have the following text:

Nunc in odio id magna molestie congue. Donec erat nulla, pulvinar eget volutpat non, molestie at nisi. Curabitur nec tristique felis. Cras imperdiet, ante et vestibulum iaculis, tellus ipsum pulvinar felis, at viverra est tellus et eros. In nec dignissim lectus, bibendum hendrerit ex. Praesent lobortis eget justo non vehicula.

Nulla et neque cursus libero tristique laoreet nec a ligula. Fusce sit “amet” scelerisque erat. Quisque lorem lectus, lobortis vitae mattis non, tincidunt sed felis. Donec sit “amet” erat nibh.

Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. In et imperdiet dui. In ut enim pharetra, blandit purus vel, malesuada est. Morbi sollicitudin eget leo nec dignissim. Praesent sed leo quis purus pretium aliquet sed quis arcu. Pellentesque facilisis tellus nulla, dignissim laoreet quam maximus et.

And I want to link the "amet" word, then I can create a shortcode:

layouts/shortcodes/amet.html

And in this shortcode I place the text:

<a href="https://www.lipsum.com/feed/html">amet</a>

In the Markdown source this is:

Nulla et neque cursus libero tristique laoreet nec a ligula. Fusce sit "{{< amet >}}" scelerisque erat. Quisque lorem lectus, lobortis vitae mattis non, tincidunt sed felis. Donec sit "{{< amet >}}" erat nibh.

Pretty easy. However Hugo by default does a little bit too much: while generating the static content, it adds a line break. Which in the browser results in:

Nulla et neque cursus libero tristique laoreet nec a ligula. Fusce sit “amet ” scelerisque erat. Quisque lorem lectus, lobortis vitae mattis non, tincidunt sed felis. Donec sit “amet ” erat nibh.
                                                                            ^                                                                                                         ^

There is a nasty little space between the word and the quotation mark. The generated HTML source shows the problem:

  <p>Nunc in odio id magna molestie congue. Donec erat nulla, pulvinar eget volutpat non, molestie at nisi. Curabitur nec tristique felis. Cras imperdiet, ante et vestibulum iaculis, tellus ipsum pulvinar felis, at viverra est tellus et eros. In nec dignissim lectus, bibendum hendrerit ex. Praesent lobortis eget justo non vehicula.</p>
<p>Nulla et neque cursus libero tristique laoreet nec a ligula. Fusce sit &ldquo;<a href="https://www.lipsum.com/feed/html">amet</a>
&rdquo; scelerisque erat. Quisque lorem lectus, lobortis vitae mattis non, tincidunt sed felis. Donec sit &ldquo;<a href="https://www.lipsum.com/feed/html">amet</a>
&rdquo; erat nibh.</p>

Usually a line break in HTML is not a big deal, the browsers deal with that. But here an unnecessary space is added while rendering the content. Hugo makes it complicated to avoid this problem. There is no default setting to turn this off, however the shortcode can have an extra Hugo tag to avoid the training linebreak:

<a href="https://www.lipsum.com/feed/html">amet</a>{{- / Strip trailing newline. / -}}

By adding the {{- -}}, the training line break is not included in the final HTML output, and the quotation mark comes right after the shortcode content.

Once or twice a year I have to setup a new Redmine project, which we use to organize conferences for PostgreSQL Europe.

Every single time I'm trying to figure out how the child pages for the meeting notes work.

Steps required:

• Create a page "MeetingMinutes", doesn't need to have any content, but it must exist
• Create the first child page with meeting minutes, as parent page select "MeetingMinutes"
• On the main wiki page insert a link with: "{{child_pages(MeetingMinutes)}}"

Every time I forget step 1), because this page is only required as parent page, and serves no further purpose. And then I'm wondering why I can't create child pages.

The Huginn software is not only good for monitoring Twitter feeds, it can also be used to monitor websites for changes.

That's actually easy and needs only two agents:

• Website Agent
• Notification Agent - Email Agent, maybe

A very common use case for hooks in git is a "pre-commit" hook. This hook is used to verify the to-be-committed data before it is added to the repository.

One important note: hooks are not part of the repository itself. Everyone can install a hook on it's own checkout of a repository, but by default the hook is not there when you clone/checkout the repository. This avoids security problems by executing arbitrary code during "git commit", or any "git" operation.
Because of this implication it is common that developers install a hook from somewhere in the repository into the ".git/hooks" directory. And in addition, the server side (the repository) can run the same checks during "git push", to enforce the rules.

Hooks in git work in a simple way: whatever program or script is run as the hook has to set a return code. If the return code is "0", git proceeds. If it's not "0", git aborts the operation.

One of the things I'm using Huginn for is monitoring for Twitter keywords. Some of them might appear in pairs, as example people like to tag Tweets about PostgreSQL with both #postgresql and #postgres. When I was using IFTTT this always created two emails, one for each hashtag. With Huginn I can deduplicate the events, and only notify about the first occurrence.

I need the following agents for this scenario:

• Twitter Search Agent
• Trigger Agent
• De Duplication Agent
• Email Agent

Huginn comes with a number of Twitter agents, but before they can be used you need to authenticate them. That process requires multiple steps, and they are not well documented.

The basics are described on the "Configuring OAuth applications" Wiki page.

For the Twitter integration one must know that you are using a single Twitter app for all users, but every Huginn user can (and should) have this app authenticated against their own Twitter account. Therefore if you want to use multiple Twitter accounts in Huginn, you need multiple Huginn users.

Huginn is a great piece of software, but the documentation is ... a bit sparse. Especially when it comes to details of the agents. I'm going to blog about a couple more more examples in the future.

For another project I'm using Leaflet, a JavaScript library for rendering maps in a browser. New versions are released occasionally, and I want to know when it's time to update the project website. Huginn can do that.