Bug 469866 - Should we "collect stats" on some bundles/fragments downloaded from repositories
Summary: Should we "collect stats" on some bundles/fragments downloaded from repositories
Status: NEW
Alias: None
Product: Platform
Classification: Eclipse Project
Component: Releng (show other bugs)
Version: 4.5   Edit
Hardware: PC Linux
: P3 enhancement (vote)
Target Milestone: ---   Edit
Assignee: Platform-Releng-Inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-06-10 12:36 EDT by David Williams CLA
Modified: 2015-06-23 08:29 EDT (History)
1 user (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description David Williams CLA 2015-06-10 12:36:22 EDT
In our promote scripts, I actually add the p2 metadata to "collect stats" on downloaded bundles (or features, or I think fragments can also be done ... any IU, basically). This is true for our Eclipse repos, as well as "Simultaneous Release" repo. 

But, so far we do not actually specify any thing to collect stats on. 

I've never thought too important, since "everyone needs the platform" :) But, I could see some advantages, to doing it for some cases -- and tracking numbers over a long period of time. 

One case that comes to mind are things like the upcoming "Java 9 Feature Patch". 

Another case that comes to mind are "fragments" for the various platforms. I suspect "launchers" would be sufficient (and assume SWT follows that same pattern, unless we think there is a huge difference between people building headless apps, vs. graphical apps?) 

And, FYI, the recommendation from Eclipse Foundation is not to do this for too many bundles/features since each "stat" requires a call back to their servers, where the "stat" is logged in the server logs. 

I'm just opening this issue to track as a possibility ... I personally have mixed thoughts about its importance ... please comment if anyone things important ... or a "waste of time". 

And, remember, it would be a "waste of time", if no one ever analyzed the data! 
Typical recommendation is to "get the numbers" on a regular basis, such as weekly, or monthly, since trends are most likely to be one of the most important aspects, not so much "grand totals". (Such as is Windows 32 bit "tapering off", and Windows 64 bit increasing, or similar, where their grand totals, say over a year long period, might be similar, and cause someone to incorrectly assume "equally popular").