1,457
edits
QA/Talkback/Metrics: Difference between revisions
→Q & A
Samuelsidler (talk | contribs) (New page: <small>« Back to QA:Talkback</small> ''Maintained by Samuel Sidler'' In the past, there haven't been useful metrics on the crashes reported v...) |
Samuelsidler (talk | contribs) (→Q & A) |
||
| Line 13: | Line 13: | ||
# '''How are you tracking "where" crashes are happening?''' | # '''How are you tracking "where" crashes are happening?''' | ||
#: Right now, we're simply doing a quick glance at the stack trace and making a best guess at where the crash is happening. If it's happening externally, we can usually see what plug-in/extension has caused it. However, if it's happening internally, we can, fairly easily, determine where it's happening and track those. This isn't being done in this first round yet, but might be integrated with Breakpad/Socorro in the future. | #: Right now, we're simply doing a quick glance at the stack trace and making a best guess at where the crash is happening. If it's happening externally, we can usually see what plug-in/extension has caused it. However, if it's happening internally, we can, fairly easily, determine where it's happening and track those. This isn't being done in this first round yet, but might be integrated with Breakpad/Socorro in the future. | ||
# '''What questions are the graphs below asking?''' | |||
#: We're asking a few questions... | |||
#:* How do we track, between releases, if we're improving or regressing behavior and causing more crashes? | |||
#:* Are we seeing more problems in our code or in third-party code? | |||
#:* If its in third party code, is it being caused by an extension, a plug-in, or something else? | |||
#:* [etc. More here.] | |||
# '''Are all topcrashes represented in your metrics?''' | # '''Are all topcrashes represented in your metrics?''' | ||
#: No. Because of the way Talkback topcrash reports are created, there are various null pointers that get tracked in very general places instead of by specific stacks. This makes them hard to track. In the future, if the reports below are useful, we can go back and generate them. However, for this first run, we're just going to skip them. This includes the first two topcrashes and several others lower in the list. | #: No. Because of the way Talkback topcrash reports are created, there are various null pointers that get tracked in very general places instead of by specific stacks. This makes them hard to track. In the future, if the reports below are useful, we can go back and generate them. However, for this first run, we're just going to skip them. This includes the first two topcrashes and several others lower in the list. | ||