I’m not sure whom I’m to file the error to. It says to “file bug report”. I was trying to access some site that I cannot remember.
The following error was shown while I was in FireFox:
Trac: Username: ron Summary: An error message request i file an error report i believe with you - sse description to An error message request i file an error report i believe with you - see description
Trac: Summary: An error message request i file an error report i believe with you - see description to Error applying Non-Tor settings: NS_ERROR_FILE_ACCESS_DENIED
Were you running your Firefox from a write-protected disk/USB key? Or perhaps as a different user than the one that owned the directory you ran it from?
Were you using TBB?
Trac: Component: Torbutton to TorBrowserButton Status: new to needs_information
The problem here is that this exception can be thrown from a wide variety of places... I can improve this message for disk full and permission denied, but other exceptions can be thrown from elsewhere, too...
Do we think it is Torbutton's job to detect these condition? What about the error messages that happen in tor's logs when it can't write to disk.. Or when Vidalia can't save configuration files.. How many warning messages do we throw at the user for these situations right now, and who should turn them into dialog boxes?
Is it a warning message or an actual failure that the user needs to do something about?
If Tor can't write to its whatever, then it cries quietly to itself, and also notifies Vidalia in case Vidalia wants to tell the user. You can find some complaints in Vidalia's message window if you happen to look there. But things generally continue working.
Do these Torbutton warnings mean that things have broken uncontrollably? If not, perhaps we should handle them as best we can and move on.
I will have to check the firefox source and/or test. I believe this exception means the prefs don't get set. We certainly stop trying to set them once we get one of these...
I can change this dialog into something more human-readable for this case and for the #3922 (closed) case easily enough, though.