[Sparkle] profiling behavior

Andy Matuschak andy at andymatuschak.org
Wed Jan 9 12:43:36 PST 2008


This definitely sounds like something I should take care of in the  
next version. Thanks for your input. I'll probably do the first  
choice, since it's useful for other things, too.

- Andy Matuschak

On Jan 9, 2008, at 12:36 PM, Mark Munz wrote:

> I've been using Tom's Sparkle-Plus for a while and one of things I've
> been looking to change is the repetitive posting of stats, which can
> skew the numbers. If I check for updates 100 times, my information
> floods the statistics.
>
> There seem to be a few options to tackle this issue:
>
> 1. send a unique ID (to prevent duplicate entries). This has some
> potential privacy issues, although if you send an ID that can't be
> traced back to a specific user, I don't think it's a huge issue.
>
> 2. only send the data if it is new or changed from the last time. This
> would check the data to be sent and compare it to the last set of data
> sent. If it changed in some way, resend it, otherwise don't. This
> could be stored in preferences which could still lead to some
> duplication, but nowhere near the send data with every request.
>
> 3. finally, only send the data every 30 days. This is a simpler
> version of #2, but would dramatically reduce duplication and would
> give a more accurate snapshot of your current active user-base by
> sampling them roughly every 30 days.
>
> Before I go forward with changes, I wanted to check to see if this was
> an issue that has already been or is planned to be handled for 2.0
> (1.5?) before I try and tackle the issue myself.
>
> I appreciate hearing your thoughts.
>
> Mark
>
> -- 
> Mark Munz
> unmarked software
> http://www.unmarked.com/
> _______________________________________________
> Sparkle mailing list
> Sparkle at lists.andymatuschak.org
> http://lists.andymatuschak.org/listinfo.cgi/sparkle-andymatuschak.org



More information about the Sparkle mailing list