Not a bad idea! The attack vector issue they mention in the PR comments is valid, though. Not displaying those errors gives an attacker no confirmation that a user whose account they're trying to attack exists, if they're trying known used passwords. But good on you doing what you can to contribute to the project!
There should be an error, but it shouldn't say whether it was the email or password that was wrong.
Fair point!
An attacker would still know the account exists, and they would know the password did not match.
_Or_ you're assuring them that specific password is used by some account, just not this one. Which is even worse.
Read the comment you're responding to, again. Nothing about their suggestion leads to either of these scenarios.
Right, I misread "shouldn't" for "should".
Couldn't they just check instancename/u/accountname to see if the user exists?
Theoretically yeah. Depends on how quickly the attacker is working or if they have enough information to know where the account lives. If they're doing their due diligence, they could 100% confirm that. But if they aren't, they might just go to a random instance, try logging in, and see if it works.
By no means is it bad to offer a response, but it always risks giving an attacker more information than they need/the victim would want to have discovered about them.
Not a bad idea! The attack vector issue they mention in the PR comments is valid, though. Not displaying those errors gives an attacker no confirmation that a user whose account they're trying to attack exists, if they're trying known used passwords. But good on you doing what you can to contribute to the project!
There should be an error, but it shouldn't say whether it was the email or password that was wrong.
Fair point!
An attacker would still know the account exists, and they would know the password did not match._Or_ you're assuring them that specific password is used by some account, just not this one. Which is even worse.Read the comment you're responding to, again. Nothing about their suggestion leads to either of these scenarios.
Right, I misread "shouldn't" for "should".
Couldn't they just check instancename/u/accountname to see if the user exists?
Theoretically yeah. Depends on how quickly the attacker is working or if they have enough information to know where the account lives. If they're doing their due diligence, they could 100% confirm that. But if they aren't, they might just go to a random instance, try logging in, and see if it works.
By no means is it bad to offer a response, but it always risks giving an attacker more information than they need/the victim would want to have discovered about them.