For those following along at home, you can now see the raw captured data @
http://mixxx.org/metrics/
Data cmetrics is collecting from my laptop (X86) and desktop (X64)...
Looks like cmetrics reports some test crap too...
__
--- == __/ t.O ==--
http://stacktrace.org/
On Wed, Jan 21, 2009 at 1:02 PM, Garth Dahlstrom <[email protected]> wrote:
> Well there are a number of reasons, the first being the original
> cmetrics server (casemetrics.net or whatever it was) was almost never
> actually up & running to actually collect any data, secondly we don't
> actually have the server code to parse or present the data from that
> server n(or any of the data that was collected, thirdly the code
> stability has been questioned in the past and fourthly there is nobody
> around to actually fix it (it probably doesn't help that the lib is
> written in C rather then C++).
>
> Those are ones I can think of. It may have been abandoned for good
> reason, but running for a bit will give us a better idea about that.
>
>
>
> On 21/01/2009, Nick Guenther <[email protected]> wrote:
> > On Wed, Jan 21, 2009 at 12:00 PM, Albert Santoni <[email protected]>
> wrote:
> >>
> >> On 21-Jan-09, at 8:12 AM, Nick Guenther wrote:
> >>
> >>> On Tue, Jan 20, 2009 at 4:03 PM, Garth Dahlstrom <[email protected]>
> wrote:
> >>>>
> >>>> I have repaired cmetrics in Mixxx such that building with "scons
> >>>> cmetrics=1" will now cause Mixxx to report back information to
> >>>> metrics.mixxx.org.
> >>>>
> >>>> The messages are being saved, unparsed into a database... I started
> >>>> writing code to parse it (since we don't have the original server-side
> >>>> implementation), but decided it is a waste of effort until we can see
> >>>> if the messages cmetrics sends are actually of any value to Mixxx
> >>>> developers or the community at large.
> >>>>
> >>>> So if there are folks out there who would mind paricipating in this
> >>>> kind of data collection you are welcome to build with cmetrics=1; I'm
> >>>> thinking we can do a dump of the dataset for people to play with at
> >>>> the end of a couple of weeks if people are interested in checking it
> >>>> out. After that we can reevaluate if cmetrics is worth having
> >>>> around/turned on.
> >>>>
> >>>
> >>> IF we have a lot of data, then cmetrics is useful as a guideline. MS
> >>> Word would have good data from such a thing. Mixxx? Probably not so
> >>> much. It's probably way more effective to just point people at the bug
> >>> tracker if they want their voice heard.
> >>>
> >>
> >> The purpose of CMetrics was two-fold:
> >>
> >> 1) To give us better demographics data and information about our users
> >> computers. Back when we were working on 1.6.0, we found that it was
> >> getting
> >> difficult to make certain design decision without understanding the
> >> hardware
> >> setups that our users have. Even in the short time that CMetrics was
> >> working, we had some good stats that helped us target development more.
> >> (As
> >> an example, the CMetrics data told us we needed to work on Hercules
> >> support
> >> more.)
> >>
> >> 2) To help us track down bugs that are otherwise impossible for us to
> >> track.
> >> Specifically, with 1.5.0, we encountered discovered a handful of "Mixxx
> >> doesn't run for me on Windows" bugs, but we had no idea that they were
> >> there
> >> because nobody was reporting them. In addition, we were unable to
> >> reproduce
> >> most of these bugs because we didn't have backtraces, etc. We knew this
> >> was
> >> still a major problem with 1.6.0 Beta 1, and now that 1.6.0 has been
> >> released, it's clear that it's STILL an issue. I don't know how to
> >> troubleshoot these bugs, because we have no data. CMetrics was able to
> >> capture stack dumps on Windows, and John had it set up so that you could
> >> load the stack dump into Visual Studio and debug Mixxx in the state that
> >> it
> >> had crashed on someone else's computer. This seemed like the holy grail
> >> for
> >> us, but our Windows development sort of died, so I'm not even sure if
> >> anyone
> >> besides John ever tried to do this. (Regardless, we're still constantly
> >> getting screwed over by these crash-at-startup bugs, so if a Win32 guru
> >> knows how we can capture _useful_ debugging data on them, please share.)
> >>
> >
> > Ah okay, good reasons. I didn't think of any of that. Well then why
> > not turn cmetrics on by default and just let the user opt-in at first
> > run?
> >
> > -Nick
> >
>
>
> --
> __
> --- == __/ t.O ==--
> http://stacktrace.org/
>
------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
Mixxx-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mixxx-devel
