Talk:Browser Metrics

From MozillaWiki
Jump to: navigation, search

Could the logs provided by this metrics project also be applied to help decide controversial usability issues (such as default prefs)? See Bug 273241 for more details on this concept. --Prognathous

The extension updateURL is not supplied for extensions hosted on A.M.O. so that won't be an option. All extensions that don't supply an updateURL are checked against A.M.O. to simplify the management of extensions on A.M.O. --Robert Strong

It's all about feedback ;)

I'd like to see a nice unobtrusive way for users to comment on specific parts of Firefox. Particular parts of the UI for example, or the extension interface.

Have the feedback sent to a central server (sorted into catagories obviously). Then run each bit of feedback through a bayesian type filter, and preform statistical analysis to find which issues mean to most to users.

Memory Leaks

I think pinpointing URLs on which we leak massively is very important to improving our end users' experience. Excessive memory use is the main complaint I get from people I've convinced to use moz. --Jonathan Watt

Extension issues

Given that Firefox is so extension dependent, a core problem-experience issue relates to stability and conflict issues, whether arising in the core or in extensions.

If Firefox does something anomalous, the common answer of "disable everything one at a time then try again" takes a long time, doesnt identify intermittents well, doesnt identify issues where 2 extensions are involved well either. This can cause users to lose faith in Firefox "I'll try something else, it may not be as good but at least it will be checked for big problems". I know I have stopped using 2 or 3 O/S programs for that reason in the past. Some users will stop using Firefox totally if they perceive that it's more trouble than it's worth to find what's causing problems/intermittents that affect their browser usage. Many users have 10-20 extensions running, so narrowing down an issue is not a short job. Many other users simply won't know how, or want to.

So given that Firefox is so extension dependent, I'd rate an improved means to rapidly and better direct users to a likely source of problem, very highly on user experience.

In practice, this means that in the event of a problem or crash or other issue, there has to be a means for Firefox to identify a likely location for it: the core, or an extension, and the relevant call or parameters involved. And the hashes and names of any extensions installed (but nothing more for confidentiality's sake)

I'd strongly be hoping for some kind of in-memory log of the last 40-60 seconds top-level program flow control, or hooked calls executed and returned from, or similar, something that if a crash or issue occurs even once, Firefox has some kind of top-level diagnostic error data which can be used to help narrow down where to look and what Firefox was doing, more than just a binary crash dump. Reconciled with privacy issues we ought at least to be able to find a way for error reports to be useful in the hands of users who can't, or won't, do debugging work for us, or whose bugs prove intermittent, unreproducible, or elusive to pin down, or for users who want to quickly identify where the bug seems to be, if they can.

I hope this is posted in the right area, it seems relevant, "information needed to be gathered to help improve user experience". Foxxen2 10:26, 1 April 2006 (PST)

Instrumentation as Extension

The current proposal is to implement instrumentation as an extension as opposed to being built in to the core product. The advantages to this approach include the ability to update the extension independently and greater flexibility to promote/market the extension. Comments on this approach are welcome and encouraged.