Closed Bug 478817 Opened 15 years ago Closed 15 years ago

Query for topcrasher list should be build from build id instead of reported date

Categories

(Socorro :: General, task)

task
Not set
normal

Tracking

(Not tracked)

RESOLVED DUPLICATE of bug 411354

People

(Reporter: whimboo, Unassigned)

References

()

Details

Currently topcrasher reports are shown for the last 2 weeks. That's not really useful. Not everyone is updating the build regularly and is always running into crashes which are already fixed since some days or even week.

We should better build up the query string against the build id. Using that way the topcrasher list isn't cluttered with all the not useful reports. See the following link. Thats a crasher which was fixed nearly a week ago but is still listed as number 1 on the list.

http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&version=Firefox%3A3.1b3pre&signature=PL_DHashTableOperate

The build id is the most useful information to be able to quickly find current top crashers. For now you have to closely inspect the graph if the given crashes still apply. That's time we could spent otherwise.

Bug 411354 could be needed to run such a query?
This can be added as part of the advanced search being worked on (i.e., you can do the search yourself when that's finished). I don't see any reason to change the way it is now because seeing trends from older builds can be useful. This also only affects development (not milestone or major) releases and would need to be implemented in such a way if we wanted to go forward (but, as I said, I don't think we do).

I vote WONTFIX (in favor of advanced search including build IDs).
In the main, for the current "pre" release, an excessive history is normally useless for me. More often that not I am looking for *recent* history, i.e. a moving history.  It doesn't normally help me to see crashes that happened 2 months ago.

That said, I'm not yet sure what makes sense here overall - long trend or short trend, and whether special casing "pre" makes sense.  perhaps a special tool would be good (asuth was discussing this on IRC), or one of the other bugs that talks about showing the change in crash frequency over time (which is what I think ss is getting at, but frankly I'm not sure topcrash for "pre" is good at that), or just ad hoc query as ss suggestions.  

Regardless, I'd rather see us think carefully on this before moving the bug to wontfix.
correction ... It doesn't normally help me to see crashes that happened [for builds that are from 1-]2 months ago.
Sam I have to second on the release builds part but for development builds we have the situation that each single build is comparable with an official release.

Eventually we should have a selection at the top where we can switch between the last 10 builds. Selecting one of those entries displays the list of crash reports for this specific build over time. So we have a way to simply compare the crash stats and how they behave, e.g when a crash has been started.

I don't wanna have to do a search when I want to walk through the top crasher list.
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → DUPLICATE
No longer depends on: 411354
Component: Socorro → General
Product: Webtools → Socorro
You need to log in before you can comment on or make changes to this bug.