Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 454927

Summary:

Some of the "detail statistics" do not make sense and are overly limited

Product:

[Eclipse Project] Platform

Reporter:

David Williams <david_williams>

Component:

Releng

Assignee:

Platform-Releng-Inbox <platform-releng-inbox>

Status:

RESOLVED DUPLICATE

QA Contact:

Severity:

normal

Priority:

Version:

4.5

Target Milestone:

---

Hardware:

OS:

Linux

Whiteboard:

Attachments:

Description	Flags
page from URL, in case URL disappears	none

Description David Williams

2014-12-11 13:09:47 EST

At least, as printed. I'll attach example, for longevity, but if the build is still available, can see 

http://download.eclipse.org/eclipse/downloads/drops4/I20141210-0800/performance/linux.gtk.x86_64/raw/Scenario1.html

Notice it has "0" for builds samples, mean and std. deviation ... I guess for what is meant to be "current build", which doesn't make sense conceptually, and which is different than the results presented right below it. 

It also has a '?' for COEF. VAR and at times I've seen "unprintable character" glyph displayed there. 

It's probably a separate bug, but it also appears to only use one run from "baseline" -- the most recent, presumably -- but, given that we run the baseline, every time we run the performance test, it seems we should take advantage of "more data" from the baseline runs ... say, a week's worth? Perhaps even have some "analysis" of baseline only data, which we'd expect to find "no change".

Comment 1 David Williams

2014-12-11 13:10:44 EST

Created attachment 249361 [details]
page from URL, in case URL disappears

Comment 2 David Williams

2014-12-12 08:46:47 EST

forgot to add this one to "umbrella bug" ... and I think bug 454987 provides a little more data.

*** This bug has been marked as a duplicate of bug 454987 ***