5 ms·

Wherein reporters don't find duplicate bugs

23 points by axiak 16y ago

redstripe 16y ago
I would call this a bug tracker fail. Having a 20,000 issues in the database is going to lead to a lot of data duplication. Searching for "mouse wheel scroll" produced 111 results. The chance that I will read through all 111 of those if I am only slightly interested in the project is: zero percent. Instead I will submit a new bug as a bit of volunteering and leave the sorting out to the maintainers. Not like one more dupe on a 20k list will make a difference. Broken window theory. At the very least they could have a prominent top 10 trending bugs list or something. It would help keep the system from being flooded when a new build introduces an easily triggered bug.
- eli 16y ago
  Really, I think the problem is that it defaults to only searching Open bugs. Makes sense, but as this case demonstrates there is a big enough gap between a bug being fixed and an updated build getting into users' hands that there's a problem. I wonder how many dupes would be eliminated if the search defaulted to Open and Recently Closed bugs.
  - aboodman 16y ago
    > I wonder how many dupes would be eliminated if the search defaulted to Open and Recently Closed bugs. As of this writing, 42 out of 259 would have been eliminated: http://code.google.com/p/chromium/issues/detail?id=66071#c217 http://code.google.com/p/chromium/issues/detail?id=66071#c21...
  - frederickcook 16y ago
    where "Recently Closed" = "Closed, but Not Yet Fully Deployed"
- barrybe 16y ago
  I've never used a bug tracker that didn't have this kind of problem, are there better ones out there? That idea of showing the top 10 trending bugs is a good one. I also like the StackOverflow approach of suggesting dupes immediately as you type stuff in, so it would be rad to have a tracker that does that.
- pnp 16y ago
  Agree on the bug tracker fail. The classification of bugs is way too primitive for a such a large project. For example, when I reported a bug, I was surprised there seemed to be no (obvious) way to search/mark bugs by release. I had no way to describe a bug as "exists in stable but not in dev release" besides adding a textual commentary. The people handling the bugs seem overwhelmed as well. A large percentage of those bugs are probably stale.
- bokchoi 16y ago
  I was pleasantly surprised by Launchpad's bug apport submission tool. When submitting a new bug from a crash, it will list some other bugs that could be related. I can either file a new bug or choose to subscribe to one of the existing bugs. Clever!
jey 16y ago
The fix is amusing: http://trac.webkit.org/changeset/73619/branches/chromium/597 http://trac.webkit.org/changeset/73619/branches/chromium/597
- nash 16y ago
  I'm betting the testing procedure is even more amusing.
gcr 16y ago
"hi, i would like to check that this issue is fixed but i'm at the bottom of that huge list of duplicates and I can't scroll up to see"

23 points by axiak 16y ago

I would call this a bug tracker fail. Having a 20,000 issues in the database is going to lead to a lot of data duplication. Searching for "mouse wheel scroll" produced 111 results. The chance that I will read through all 111 of those if I am only slightly interested in the project is: zero percent. Instead I will submit a new bug as a bit of volunteering and leave the sorting out to the maintainers. Not like one more dupe on a 20k list will make a difference. Broken window theory. At the very least they could have a prominent top 10 trending bugs list or something. It would help keep the system from being flooded when a new build introduces an easily triggered bug.

eli 16y ago
Really, I think the problem is that it defaults to only searching Open bugs. Makes sense, but as this case demonstrates there is a big enough gap between a bug being fixed and an updated build getting into users' hands that there's a problem. I wonder how many dupes would be eliminated if the search defaulted to Open and Recently Closed bugs.
- aboodman 16y ago
  > I wonder how many dupes would be eliminated if the search defaulted to Open and Recently Closed bugs. As of this writing, 42 out of 259 would have been eliminated: http://code.google.com/p/chromium/issues/detail?id=66071#c217 http://code.google.com/p/chromium/issues/detail?id=66071#c21...
- frederickcook 16y ago
  where "Recently Closed" = "Closed, but Not Yet Fully Deployed"
barrybe 16y ago
I've never used a bug tracker that didn't have this kind of problem, are there better ones out there? That idea of showing the top 10 trending bugs is a good one. I also like the StackOverflow approach of suggesting dupes immediately as you type stuff in, so it would be rad to have a tracker that does that.
pnp 16y ago
Agree on the bug tracker fail. The classification of bugs is way too primitive for a such a large project. For example, when I reported a bug, I was surprised there seemed to be no (obvious) way to search/mark bugs by release. I had no way to describe a bug as "exists in stable but not in dev release" besides adding a textual commentary. The people handling the bugs seem overwhelmed as well. A large percentage of those bugs are probably stale.
bokchoi 16y ago
I was pleasantly surprised by Launchpad's bug apport submission tool. When submitting a new bug from a crash, it will list some other bugs that could be related. I can either file a new bug or choose to subscribe to one of the existing bugs. Clever!

jey 16y ago

The fix is amusing: http://trac.webkit.org/changeset/73619/branches/chromium/597 http://trac.webkit.org/changeset/73619/branches/chromium/597

nash 16y ago
I'm betting the testing procedure is even more amusing.

gcr 16y ago

"hi, i would like to check that this issue is fixed but i'm at the bottom of that huge list of duplicates and I can't scroll up to see"