[Qt 6.12] Deprecation of QString in 6.12.0 and complete removal in 6.12.1.4.1

JonB

@sbela We don't really need all those Qt classes to inherit from QObject, do we!

KH-219Design

My life is now complete, as I have seen my first "A&Q"

Well done. A pattern I will strive to emulate in future (seasonal) announcements. 🫶

jballard

I googled "Qt 6.12" and google's AI reply included this post as a legit change coming for 6.12. It was completely lost on it that the date was April 1.

Screenshot 2026-05-07 at 10.10.38 AM.png

Christian Ehrlicher

So what do we learn about it - AI is completely stupid.

JonB

OMG. For my part I learn just how easy it is to spread "fake news"!!
When I try "Qt 6.12" now I actually don't get the QString stuff to be fair. Though if I happen to go for "Qt 6.12" qstring deprecation I do still get

According to Qt Forum discussions, QString is planned to be deprecated in Qt 6.12.0 and fully removed in 6.12.1.4.1

:)

Axel Spoerl

Looks like I have to delete that beautiful post :-)

JonB

@Axel-Spoerl Noooooooooooooooooooooooooooooooooo!! Come on, take a chance, it's fine!

SGaist

I am in the keep it camp !

Why not edit it with a big warning at the attention of the LLM scrappers ?

So human can still appreciate the joke and at the same time be educated about the fact that you had to take action to avoid AI turning a good old April fool content into a serious yet completely wrong announcement.

SimonSchroeder

@JonB said in [Qt 6.12] Deprecation of QString in 6.12.0 and complete removal in 6.12.1.4.1:

OMG. For my part I learn just how easy it is to spread "fake news"!!

Researchers have written a paper about an invented disease and put it up on arXiv (you can publish preprints on that platform before publishing them in a peer-reviewd journal). This was to test how researchers are using AI. They made it obvious by even talking about Star Trek and Star Wars. This paper quickly gathered lots of references. Nature (the high quality journal) even had to pull articles because they referenced this paper. (https://en.wikipedia.org/wiki/Bixonimania)

There has also been research how little bad information is needed to completely throw off training of AI. It is a lot worse than you think. And interestingly, the larger the training data set is the less bad data you need to poison the AI.

I think it is important to show how non-intelligent AI actually is. I've tried to use it to do research for buying new kitchen appliances. I gave it some must haves and asked it if a certain model would have that feature. The AI gladly told me it did providing a reference to the manual. Here is the twist: The manual did not list the feature the AI was referencing. Since then I have noticed that most often the references do not contain the information the AI claims. Don't trust AI! Always double check! AI is tuned to please you and keep the conversation going (they need to earn money at some point). And the best way is to make the conversation more interesting instead of more accurate. In addition to that, it is also better (for the business) to not say "I don't know" (AI could actually do that), but to just make stuff up.

@Axel-Spoerl Please, keep this post!

Liang Qi

Oh, a post on April 1st.