Researching fuzzy stack hashing via Apport and Launchpad
Hello,
I'm taking a quick look at the efficacy of fuzzy stack hashing by inspecting how well Apport is automatically triaging crashes via the Launchpad bug database. I have a simple heuristic that seems to be working for finding Apport-triaged crash reports in Launchpad. I simply look for bugs whose title matches the Apport format string for crashes ("X crashed with Y") and then look for some of the usual Apport-generated attachments. So I can count the number of crashes. I'd like to figure out how many duplicates Apport generated (i.e., the cases where Apport failed). I can count the number of duplicate reports that were submitted, but I can't be sure if the duplicate reports were also Apport-triaged crashes since duplicate reports seem to be deleted.
So here is my question: Is there a way to get at the seemingly deleted duplicate reports, and if not, do you guys have an idea if duplicate crash reports are usually also Apport-reported?
This whole topic may seem extremely dry, but I'm of the mentality that we can't be sure we are improving on fuzzy stack hashing without first measuring a baseline.
Thanks in advance for any comments. If you have any questions just let me know.
JF
Question information
- Language:
- English Edit question
- Status:
- Expired
- For:
- Apport Edit question
- Assignee:
- No assignee Edit question
- Last query:
- Last reply: