Hi, I 100% agree with Kai's original suggestion, and it has been my thought as well for Qt 7: We should get rid of AutoText completely. The convenience of not having to set an explicit text format (one line of code) is debatable, and in my opinion nowhere near worth the risk of text fields accidentally being exposed to network access. Accidental rich text is always a bug, so it should always be explicit.
Given that AutoText is primarily a compatibility option at this point, I don't think we should do anything to mightBeRichText() that might change existing behavior (or add overhead). It is what it is. However, I do think we should also consider ways of changing the default in Qt 6 lifespan. We have to prioritize backwards compatibility, so we can't change it unconditionally for everyone. My idea so far has been that we could add application profiles which govern how defaults and behavior are chosen. With the default "compatibility" profile, backwards compatibility would be the priority like before. But for applications explicitly choosing a "security" profile, we would be able to change defaults and behavior between releases as long as we think it strengthens the security of the application by avoiding common pitfalls. Security-sensitive applications would pick this profile and accept the extra price of sometimes having to update their code. I might make a session for this in the Contributor Summit, actually. It would be nice to hear if there are other ideas. -- Eskil Abrahamsen Blomfeldt Principal Manager, Graphics Engines The Qt Company Sandakerveien 116 0484 Oslo, Norway [email protected] http://qt.io Public ________________________________ From: Development <[email protected]> on behalf of Edward Welbourne via Development <[email protected]> Sent: Wednesday, July 29, 2026 10:13 AM To: Kai Uwe Broulik <[email protected]>; [email protected] <[email protected]> Subject: Re: [Development] Proposal: Disable AutoText in Qt 7 Kai Uwe Broulik (29 July 2026 09:52) wrote: > Qt has a feature called AutoText where when you have a label (both in Qt > Widgets and Qt Quick) and its contents “look like HTML” > (Qt::mightBeRichText) it will switch to rich text (styled text) and > start interpreting it as HTML. > > Just name a file or a Wifi SSID "<h1>hello.txt" and ... A stray thought prompted by this: Qt::mightBeRichText() uses a fairly dumb heuristic, just looking for something tag-like before the first newline. One potential mitigation would be to relegate that to a pre-test and have a slightly more stringent test for AutoText, say looksLikeRichText(), which uses that precheck and, if it passes, does something closer to a (cheap as we can sensibly make it) check that the apparent rich-text is well-formed. For plain text this stays cheap, for HTML it costs a bit more, but we'd avoid the "mark-up spills out of the text into other things" issue (assuming we made that the focus of the approximate-validity check). I'm not familiar with the relevant infrastructure, though, so perhaps that increase in cost would be a significant impediment to making such a change. If not, would it in fact mitigate the problem you're describing adequately ? Eddy. -- Development mailing list [email protected] https://lists.qt-project.org/listinfo/development
-- Development mailing list [email protected] https://lists.qt-project.org/listinfo/development
