This site was/is a gem, I used it many-a-times as my once a week reference for regex pre-AI.
I also adore debuggex.com which still seems to be working. Debuggex UX is not as nice, but it does give you a visual representation of your regex which is a nice touch.
Shout out to self-hosted alternatives like https://github.com/SunneV/OpenRegex
I hope this is preemptive instead of a reaction to something.
Automated probing of the web is only going to get worse from here on. It is going to make SaaS more expensive as there will need to be actual ongoing maintenance instead of the ‘build it once run it forever’ ideological thinking of software business models.
Maybe we will see renewed interest in publish once desktop apps as they will have limited risk?
I doubt we will, people want to be able to see the same documents/messages/etc from their PC and their phone.
That doesn't strictly require a central server, but a central server certainly makes any kind of sync orders of magnitude easier for everyone. That becomes a central weakness whether the client software is deployed via javascript in a web browser or packaged as a desktop app.
Running a proprietary service might buy you a small amount of time compared to web-scanners, but it doesn't change the risk model. In fact, given that the browser is still a fairly effective sandbox, moving to published desktop app probably increases the risk for clients by increasing the blast radius of any issues there.
Consider the kind of viruses we used to deal with: https://en.wikipedia.org/wiki/Melissa_(computer_virus)
Depends if said apps attempt to load untrusted data like files.
What data could they have that needs to be secured?
EDIT: my apologies, I did not realize this used server-side parsing
People probably put a fair amount of arguably-secret information into various “calculator” sites.
As such, I can imagine how testing out a regex could leak information that might be interesting to someone.
https://github.com/gskinner/regexr/blob/3.6.1/server/schema.... and https://github.com/gskinner/regexr/blob/3.6.1/index.php imply the db is probably full of `/AKIA[A-Z0-9+]/.test("AKIA00...")` type literals and they are presumably worried about the stack coughing up access (not to mention https://github.com/gskinner/regexr/blob/3.6.1/server/Config.... but the maintenance page seems to imply they are more worried about the stored data than their own creds)
usernames and password hashes?
I heard they were the custodians of a mythical, dangerous regex that could parse HTML.
Not sure what RegExr is built with (never used it), but lots of the recent Ruby vulnerabilities are exploitable with regexes based on user input.