Uncategorized

LinkedIn and AI Training

It’s not quite a week since I first saw the meme announcing that LinkedIn had just become the latest high-profile company to start using their users’ data to train an AI system, and (of course) it was opt-out. So if you did nothing (e.g. you didn’t see the notification) then they would simply assume you were OK with it.

Initially, you could just go to https://www.linkedin.com/mypreferences/d/settings/data-for-ai-improvement to turn off the setting, but clicking that link doesn’t seem to be reliable. (If I go directly to that link, it works, but if it goes through Facebook’s link tracker, it goes to a page not found error.)

In a desktop web browser, you can click on “View Settings” in the left navigation, click on “Data Privacy” and then, under the “How LinkedIn uses your data” heading, click on “Data for Generative AI Improvement.” At the moment, that page has a single toggle for “Use my data for training content creation AI models.”

I don’t have the app installed myself, but I’m told you can similarly go to Settings -> Data Privacy -> How LinkedIn Uses Your Data -> Data for Generative AI Improvement

You can read more about this in The Washington Post’s Tech Friend column (no account required, this is a gift link)

Tip of the hat to John Newmark who first made me aware of this.

Uncategorized

Turn Off The Dictation Hotkey

I have a habit where I’m about to type a control key (e.g. control-c), I’ll hit the control key twice. I don’t know why I do this, and I’m not sure why it’s only the control key, but on Mac, this has the unwanted side-effect of popping up a prompt to enable dictation. (If I had dictation enabled, I suspect it would start transcribing my speech, which might be even worse.)

Macintosh dialog box asking, "Do you want to enable Dictation."

To turn this off:

Go to system preferences and scroll down to “Keyboard.”

Macintosh preferences dialog with the "Keyboard" pane active.

At the bottom of the keyboard pane, is the “Shortcut” label. This shows the current hotkey for activating dictation. I really don’t want this hotkey, but there doesn’t seem to be an option to not have one, so I’ll choose the microphone and hope it doesn’t start transcribing every random conversation near the MacBook.

List of dictation hotkey options: "Press Microphone," "Press Control Key Twice," "Press Globe Twice," "Press Right Command Key Twice," "Press Left Command Key Twice," "Press Either Command Key Twice," "Customize..."

(Cover image generated by AI via bing.com/create.)

Uncategorized

Installing PHP modules

I’m running a WordPress instance on a Zimaboard on the home network. Normally, that’s the kind of thing I’d put on the paid hosting service so I can let someone else worry about patching the OS and such, but since family calendars and the like don’t need to be on the public internet, I decided to do this one in-house.

Once I got WordPress up and running, I checked the “Site Health” and along with some old themes that needed to be cleaned up, there was a notice of a critical issue, telling me that “One or more required modules are missing.”

Expanding the dropdown, I saw the list included curl, imagick, zip, and gd. I tried to be thorough when installing PHP, but evidently missed a few. No biggie. Here’s how to fix it.

Performance warning that "One or more required modules are missing" PHP modules perform most of the tasks on the server that make your site run. Any changes to these must be made by your server administrator. The WordPress Hosting Team maintains a list of those modules, both recommended and required, in the team handbook. Warning: The optional module, curl, is not installed, or has been disabled. Warning: The optional module, imagick, is not installed, or has been disabled. Warning: The optional module, zip, is not installed, or has been disabled. Error: The required module, gd, is not installed, or has been disabled.

This is all running on Ubuntu, so the first step is to update the list of available packages. Because you always do that first.

$ sudo apt update -y

Next, install the missing packages. These are php modules, so the package to install is named “php-” and then the module name. (e.g. php-curl).

$ sudo apt install php-gd php-imagick php-curl php-zip

My machine is running PHP 8.1, so apt determined that the correct packages to install were php8.1-gd, php8.1-imagick, php8.1-curl, and php8.1-zip

Now I know the php-imagick module depends on ImageMagick, so I wanted to make sure that was installed, so after checking the apt command’s help text, I ran apt list -a ImageMagick

$ apt list -a ImageMagick
Listing... Done
imagemagick/jammy-updates,jammy-security,now 8:6.9.11.60+dfsg-1.3ubuntu0.22.04.3 amd64 [installed]
imagemagick/jammy 8:6.9.11.60+dfsg-1.3build2 amd64

Excellent. Everything should be good now right? That’s what I thought, but WordPress disagreed. Returning the Site Health page, the same message appeared, telling me that “One or more required modules are missing.”

I had to scratch my head for a bit on that one. Then I remembered, you don’t just install PHP modules, you also have to tell PHP you want to use them. (For example, Xdebug notoriously causes programs to run more slowly.)

You use phpenmod to enable modules and phpdismod to disable them. (Arguably, I should have included “-s apache2” so they would only be enabled for Apache, but I wanted to make them available for any use of PHP.)

After enabling the modules, you also need to restart apache so the new modules are loaded, so the final set of commands is:

$ sudo phpenmod curl imagick zip gd
$ sudo systemctl restart apache2

And now the Site Health report is a little bit happier.

Uncategorized

Set up a MySQL Database for WordPress

I keep losing track of the file where I have these steps written down. It’s past time to put them someplace where I can find them, and perhaps help a few others as well.

create database SOME_DATABASE;
create user 'SOME_USER'@'localhost' IDENTIFIED BY 'A_STRONG_PASSWORD';
GRANT ALL ON SOME_DATABASE.* TO 'SOME_USER'@'localhost';

The database permissions grant can be fine-tuned a bit, e.g. after installation, remove the DROP, ALTER, and GRANT permissions. (This does, of course, depend on what your plugins are doing, and potentially the needs of a particular major version upgrade.)

tech

Setting defaults for the dig command

Today I learned you can set default output options for the dig command by creating a .digrc file in your home directory.

Ordinally, running the command dig www.chaosandpenguins.com, the result is this rather hefty block of text.

$ dig www.chaosandpenguins.com

; <<>> DiG 9.16.1-Ubuntu <<>> www.chaosandpenguins.com
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 40732
;; flags: qr rd ad; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 0
;; WARNING: recursion requested but not available

;; QUESTION SECTION:
;www.chaosandpenguins.com.      IN      A

;; ANSWER SECTION:
www.chaosandpenguins.com. 0     IN      CNAME   chaosandpenguins.com.
chaosandpenguins.com.   0       IN      A       216.92.152.175

;; Query time: 0 msec
;; SERVER: 172.28.224.1#53(172.28.224.1)
;; WHEN: Wed Nov 16 23:13:00 EST 2022
;; MSG SIZE  rcvd: 136

That’s a whole lot of text. So let’s add a couple options. +noall turns off everything. Running dig www.chaosandpenguins.com +noall would literally return nothing at all. To bring back the answer section (which is what I’m interested in most of the time), you add the +answer option.

$ dig www.chaosandpenguins.com +noall +answer
www.chaosandpenguins.com. 0     IN      CNAME   chaosandpenguins.com.
chaosandpenguins.com.   0       IN      A       216.92.152.175

That’s much more compact , but getting it requires some extra typing. And I want that version of the output most of the time, so wouldn’t it be nice if there was to make that the default?

This is where the .digrc file comes in. You create it in your home directory and just put in a single line containing the options you want. So, to make +noall +answer the defaults, I just run this command:

$ echo +noall +answer > ~/.digrc

And now when I run dig www.chaosandpenguins.com without any options, here’s the default output:

$ dig www.chaosandpenguins.com
www.chaosandpenguins.com. 0 IN CNAME chaosandpenguins.com.
chaosandpenguins.com. 0 IN A 216.92.152.175
Uncategorized

Getting Verified on Mastodon

2 Comments

I’m not leaving Twitter, not yet anyhow, but over the weekend I decided to dip my toes in the water and check out Mastodon. I’m still collecting my thoughts on the topic (more on that later perhaps), but in the meantime, let’s talk about how to get verified.

Last week, Twitter announced you’d soon be able to get verified (i.e. a blue checkmark confirming that you’re who you say you are) by paying eight dollars a month. I think they’ve missed the boat.

For me, verification isn’t “Oh, this person’s name really is ‘Roy Kent.” It’s about verifying the account belongs to the same Roy Kent who used to play for AFC Richmond and not someone who coincidentally has the same name and uploaded the other guy’s publicity photo. (Update November 13, 2022: Oh wow! I thought they’d at least check that the name matched what was on the credit card. Instead, the blue check mark has gone from “This person is who they claim to be” to something more akin to the The Star-bellied Sneetches.)

Being a collection of independent web sites, verification on Mastodon doesn’t work quite the same way. It’s quite possible (and arguably, desirable) that @thatblairguy@mstdn.social and thatblairguy@mastodon.social (or any other server) are different people.

The “tricky” part is, you need your online identity to be more than “I’m this person on this particular social media site.” (If your identity really is, “I’m this person on this particular social media site,” I don’t understand the value of verification – if you’re only on one site, where else do people know you from?)

If you do have a web site (or really, any page you control), you can link to it from your Mastodon profile. Then, while on the “Edit your profile” page, copy the verification link and put that on the page you linked to. This causes Mastodon to display the link to that page in green, meaning “verified.” (Or, put another way, “The person controlling that other site is also the person who controls this profile.”)

Using my own profile as an example, here’s how to do it (some of the details may vary between servers, but I believe the general steps will be the same):

Step 1: Edit your profile.

Step 2: Scroll down to the “Profile metadata” section. And put in both a label and a link to your website. (I labeled mine as “website” but you can use whatever text you want.)

Step 3: Click “Save Changes”. (This saves your work, but you’re not verified yet.)

Profile metadata block listing thatblairguy.com and chaosandpenguins.com as web sites I control.

At this point, your mastodon profile will show a link to your web site, but it won’t display that you’ve verified the ownership. You might still be some rando trying to claim that you own someone else’s web site. Next, you have to prove that you own the page you’ve linked to.

Step 4: Scroll back down to the “Profile metadata” section and click the “Copy” button under the verification text. This puts a link into your paste buffer, for my account, it looks like this:

<a rel="me" href="https://mstdn.social/@ThatBlairGuy">Mastodon</a>

Place that link somewhere on the page and after a short while, your profile page will show the link to your web site with a highlight, verifying that you’re the same person who controls that page.

Do bear in mind that it might take a few minutes for the highlight to show up. Most mastodon servers are being run by volunteers and at the moment, they’re under a heavy load with people moving from Twitter.

Profile showing thatblairguy.com and chaosandpenguins.com as verified.

Oh, and if you really, really want a checkmark next to your name, you can add an emoji as part of your display name. ✅

Uncategorized

Immutable Objects in JavaScript

Yesterday I learned you can have immutable objects in JavaScript.

Constant values of course are nothing new (e.g. const PI = 3.14), at least for simple values. But for objects, the name of the constant only affects which object the name refers to, not the value of the object’s members.

const a = { num: 5}
console.log(a.num);
// outputs: 5

a.num = 10;
console.log(a.num);
// outputs: 10

a = {}
// TypeError: Assignment to constant variable.

So it turns out you can make an object immutable by passing it to Object.freeze().

So if we take the previous example and freeze the object, changing the property’s value doesn’t work.

const a = { num: 5}
console.log(a.num);
// outputs: 5

Object.freeze(a);

a.num = 10;
console.log(a.num);
// outputs: 5

The thing that bothers me about this is the attempted assignment fails silently. That opens the possibility of some hard to find bugs.

Start the code with 'use strict', and now there’s an explicit runtime error.

'use strict'
const a = { num: 5}
console.log(a.num);
// outputs: 5

Object.freeze(a);

a.num = 10;
// TypeError: Cannot assign to read only property 'num' of object

The other thing to note is that, much like const does for variables, freeze()only affects the values of simple properties. If a property refers to an object, that object’s properties won’t be frozen.

The examples in the Mozilla documentation includes a discussion of “shallow freezing” a deepFreeze() example method demonstrating a recursive technique for freezing object members recursively.

(Cover image via WikiMedia commons, used under Creative Commons Attribution-ShareAlike 1.0 Generic)

Uncategorized

Turn off the “finish setting up your device” screen

One of my peeves with Windows 10 is occasionally I’ll get a screen saying “Let’s finish setting up your device.” (Uh, I finished setting it up a couple years ago, why do you keep suggesting this?)

This evening, I spotted a post on Twitter from @MrTurner asking how to get rid of that prompt. Great question! And there was an equally awesome reply from @Lucas_Trz with the answer.

So, just in case that isn’t clear enough.

  1. Go into settings (click the Start button, and then the “gear” icon), click 
  2. Click “System”
  3. Click “Notifications & actions”
  4. Uncheck the box next to the text that starts off with “Show me the Windows welcome experience after updates and occasionally” (I’d like to suggest you might want to uncheck a few other things as well.)

Going forward, this is on my list of things I’ll do any time I reinstall Windows. Right next to turning off the feedback surveys.

Cover image by twitter user @MrTurnerj, used in the context of a critique of Windows.

Uncategorized

Why your organization needs to own its email address

Dear hypothetical reader: this is a coalescence of some thoughts that have been circulating through my head over the past several years. Hopefully if I write them down, I can make room for other, more interesting thoughts.

I’m opposed to the idea that an organization (I apply this equally to civic organizations, club, and businesses) might use email addresses which the people doing the organization’s business set up with their personal email provider of choice.

That is to say, the organization’s leadership shouldn’t send email to the general public (or even the organization’s own members) from @yahoo.com, @gmail.com, or @whatever email address that isn’t owned by the organization. (Ideally, this would be the same domain name as the organization’s web site, but I do recognize that for some small businesses, their primary web presence is Etsy or something similar.)

I have three main reasons for this

  • It doesn’t look professional. People expect to get email from the organization they’re transacting with. (Would you do business with Amazon if the emails came from JeffyB64@yahoo.com?)
  • Having mail boxes the organization administers provides a fallback if someone forgets their password. (This was recently driven home for me when a friend lost access to 20+ years of business correspondence because of a lost AOL password.)
  • Having email addresses the organization administers protects both the organization and the individual if someone leaves under less than amicable circumstances. (Someone leaving under such circumstances is unlikely to be happy if asked to forward emails indefinitely.)

(Image via Pixabay user Deans_icons used under Pixabay License.)

tech

Making a “dumb” lamp smart with Home Assistant

I have a parrot – her name’s Terry Datyl. I also have a bunch of house plants. These two statements are connected, though perhaps not in an obvious manner.

Part of my morning routine is to go downstairs and uncover Terry’s cage, turn on the light next to the cage, and then turn on the radio so she has a “flock” of sorts to hang out with. Next, I turn on the various grow lights for the house plants.

That was the routine for several years. I’d connect lamp timers when I was gone for the weekend and it all worked fairly well.

A while back, I thought it might be interesting to have Home Assistant take over turning on the plant lights. Ideally, a trigger would fire when Terry’s lamp was turned on and I could just put some smart plugs on the plant lights.

Have you ever tried shopping for a smart lamp? They do exist, but it mostly seems like accent lighting. I haven’t found anything which was meant to illuminate a room and nothing that really fit into our décor anyhow.

So, what I hit on was the idea of connecting the existing lamp (the “control lamp” if you will) to a TP-Link HS110 smart plug. In addition to the vanilla on/off capability, this particular plug also reports power usage. (Note: This may no longer be possible with TP-Link smart plug. A 2020 firmware update removed an API Home Assistant relies on. The general concepts however should still apply to other smart plugs with power monitoring capabilities.)

So, step one was to connect the plant lights to smart plugs. I used TP-Link plugs because I already had them, but you can use others. (As noted above, that may indeed be necessary.)

Next, I created two “scenes” in Home Assistant. One in which all the plant lights were on, and another where they were all off. (Creatively named, “Plants Off” and “Plants On.”)

Next up was setting up the HS110 plug that would be running the show. In addition to an on/off state, the HS110 exposes several state “attributes”: voltage, current, and the current (meaning “right now”) power in watts. This last one is what we’re interested in.

State attributes vary from one device to the next, so Home Assistant doesn’t really have a good way to expose them directly. Instead, you can expose the values you want via sensor templates.

Go into the Home Assistant config directory and edit the sensor.yaml file (creating it if necessary). Here’s the entry I created to read how many watts Terry’s lamp is using

- platform: template
  sensors:
    terry_s_lamp_watts:
      friendly_name_template: "{{ states.switch.terry_s_lamp.name}} Current Consumption"
      value_template: '{{ states.switch.terry_s_lamp.attributes["current_power_w"] | float }}'
      unit_of_measurement: 'W'

Note: the friendly_name_template and value_template entries are one line apiece (one day I’ll tweak this theme to better accommodate code snippets). This is just a plain YAML file so all the usual editing concerns apply.

This shows up in Home Assistant as a numeric sensor with the name sensor.terry_s_lamp_watts (you may need to restart for it to show up). It’s a floating-point number, so we’ll have to accommodate for that in the automation.

You’ll need to know how much power the lamp draws in its on and off states so in the Home Assistant UI, go to Developer Tools > States and select the entity sensor.terry_s_lamp_watts (substituting your sensor’s name, of course) and if all is right, the current power draw will show up in the “State” field. Turn the lamp on and off (using the the lamp’s switch, not the smart switch) and note the values (click the “refresh” icon to get the new value).

The final step then is to create two automations. Mine are named “Terry’s lamp turns on” and “Terry’s lamp turns off.”

For the “on” automation, I used a numeric state trigger on the entity sensor.terry_s_lamp_watts. The power usage tends to vary over time as the bulb warms and cools, so I chose 9 watts as a value that comfortably below the lamp’s “on” state while still higher than the lamp’s “off” state. (Similarly, the “off” automation uses a value of 5 watts, which allowed me to turn on the radio that was plugged into the same smart plug without triggering the automations.)

For both automations, the only action is to activate the appropriate scene. Either scene.plants_on or scene.plants_off.

At this point, you now have a single lamp which uses its existing switch to control other lights. (This means, no worries about guests messing things up by not using the smart switch – this adds smarts to the “dumb” lamp.)

Latency

With the HS110 smart plug on the control lamp, there’s a delay of up to 30 seconds between the time the control lamp changes state and when the automation will run. That’s because TP-Link smart plugs don’t actively report their state and Home Assistant has to use “local polling” to check whether the switch’s state has changed since the last time it was set. In order avoid flooding the local network with traffic, Home Assistant only checks the device’s status once every 30 seconds. Devices from other manufacturers may behave differently.

To solve this using the HS110 device, I added an additional automation, using a trigger type of “Time Pattern” with Hour and Minutes left blank and Seconds set to /3, the automation runs every 3 seconds. I then set the action manually, in the UI’s YAML editor

service: homeassistant.update_entity
data: {}
entity_id:
  - switch.terry_s_lamp
  - switch.office_light

This action causes Home Assistant to update the state for both the switch.terry_s_lamp (the control lamp) and switch.office_light (the smart switch in my home office, also a TP-Link device).