[dibes]

This would be particularly useful when asking specifics about when the error arose. It would also be nice if there was an option to attach a censored client screenshot (scrub username if possible) of what happened when the exception arose.

After brief discussion in the posts below, here are some possible other implementations that would be preferable to some:

Exception communication

1. The user has the option to provide what the bot was doing at the time if they know in an input box to provide more context for the error.

Screenshot

1. The screenshot is taken on the event that an exception happens
2. The username is immediately censored out in all interfaces where it is visible
3. The user then has the option whether or not they want to send the screenshot at all
4. If the user agrees, the image pops up in a simple editor in which they can further censor information that could be sensitive.

 

[derk]

To whom would you respond? Cloud/Arbiter or the the user who experienced the error?

And I think screenshotting would cause a lot of security issues.

 

[savior]

I don't support this, the exception log shouldn't be a platform of communication. If there is any bug which might be discussed further it should be discussed in the corresponding thread.

 

[dibes]

To whom would you respond? Cloud/Arbiter or the the user who experienced the error?

And I think screenshotting would cause a lot of security issues.

To the user in an anonymous thread. And the screenshot would be opt in by the user and scrubbed of their username (you could query the interfaces and find the X, Y loc of the usernames and censor it).

 

[derk]

To the user in an anonymous thread. And the screenshot would be opt in by the user and scrubbed of their username (you could query the interfaces and find the X, Y loc of the usernames and censor it).

The username isn't the only thing that could contain information about the identity of the user. With the availability for EVERYONE to make scripts, and maybe adding errors on purpose, can gain information about accounts.

 

[dibes]

I don't support this, the exception log shouldn't be a platform of communication. If there is any bug which might be discussed further it should be discussed in the corresponding thread.

My main reason to open a thread on communication on a specific log is that it can provide better context to the exception at hand. I understand your point and in which I think a compromise could be made so that instead of a thread of communication the user should just be able to submit context in which the bug happened. So just an input box where they can describe what was happening if they know.
 

The username isn't the only thing that could contain information about the identity of the user. With the availability for EVERYONE to make scripts, and maybe adding errors on purpose, can gain information about accounts.

Hmm, I see what you mean. Perhaps the user could censor their own screenshot after it scrubs their username so they can hide sensitive information. That would be on top of making is 100% opt in to send the screenshot in the first place.

 

[derk]

My main reason to open a thread on communication on a specific log is that it can provide better context to the exception at hand. I understand your point and in which I think a compromise could be made so that instead of a thread of communication the user should just be able to submit context in which the bug happened. So just an input box where they can describe what was happening if they know.
 


Hmm, I see what you mean. Perhaps the user could censor their own screenshot after it scrubs their username so they can hide sensitive information. That would be on top of making is 100% opt in to send the screenshot in the first place.

That's a decent idea. Then again if users aren't forced to make a screenshot I don't see a lot of people actually doing it.

 

[dibes]

That's a decent idea. Then again if users aren't forced to make a screenshot I don't see a lot of people actually doing it.

I guess it depends on what actually happens. In my opinion it would be worth trying to make it opt in and see if people use it and if it is actually useful for the developers. If it is useful but no one really uses it perhaps arbiter and cloud can deliberate whether they want to make it required.

 

[arbiter]

Let's first do stack trace obfuscation reversal and then re-evaluate whether this is still desired. In my experience, I can safely ignore user feedback and debug exclusively from the stack trace.