[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Fw: [platform-releng-dev] [eclipse-build]Build I20050412-1547 (Timestamp: 200504121547):Automated JUnit testing complete. Test failures/errors occurred.
|
I got an explication on these abnormal results...
I was running performance tests with a 3.1 M6 SDK and just
org.eclipse.jdt.core project in my workspace (synchronized with HEAD).
Refreshing this eclipse install to last build I20050412-1547,
I now retrieve performance results displayed for this build...
So, as with same JDT/Core version (I used v_550 to be sure) I get
different performance results on DOM>Creation and Search>Indexing
tests, it sounds like there were some changes in prereq plugin(s)
which slowed down JDT/Core in DOM/AST and Indexing area...
I've also installed last week build (I20050405-0800) and results
are similar to M6 one.
So, problem has been introduced since last week. I continue to
investigate and try to identify where this gap comes from.
Cordialement/Regards,
Frédéric
----- Forwarded by Frederic Fusier/France/IBM on 04/13/2005 03:53 PM -----
Frederic
Fusier/France/IBM
To
04/13/2005 02:06 "Eclipse platform release
PM engineering list."
<platform-releng-dev@xxxxxxxxxxx>
cc
Subject
Re: [platform-releng-dev]
[eclipse-build]Build I20050412-1547
(Timestamp: 200504121547):Automated
JUnit testing complete. Test
failures/errors occurred.(Document
link: Frédéric Fusier)
I'm really surprised by org.eclipse.jdt.core performance results of this
integration build.
I cannot reproduce deviations locally for "DOM>Creation" and
"Search>Indexing"tests .
On my Linux box reserved to performance tests, I got around +1.23% (ie.
green) for
"DOM>Creation" tests and +2.37% (ie. still green) for "Search>Indexing"
one.
These results are computed after having run tests 10 times both on 3.0 and
3.1 streams.
As standard deviation of these results is around 1%, I'm really fully
confident on their accuracy.
So, I wonder if there could be a problem on test machines (3.0 reference
not rebuilt, issue while running
performance test, etc...) which can explain these so big differences?
However, I've noticed that one run (over 10 executed) of "Search>Indexing"
test
got an abnormal delta of -23% (ie. red)...
So, I'm currently investigating to understand how this was possible and
if test procedure can be responsible in this kind of result
(typically unexpected garbage collect...)
Cordialement/Regards,
Frédéric
eclipse-releng@ec
lipse.org
Sent by: To
platform-releng-d platform-releng-dev@xxxxxxxxxxx
ev-bounces@eclips cc
e.org
Subject
[platform-releng-dev]
04/13/2005 03:04 [eclipse-build]Build I20050412-1547
AM (Timestamp: 200504121547):Automated
JUnit testing complete. Test
failures/errors occurred.
Please respond to
"Eclipse platform
release
engineering
list."
Build I20050412-1547 (Timestamp: 200504121547): Automated JUnit testing is
complete. Test failures/errors occurred in the following:
org.eclipse.core.tests_linux.gtk
org.eclipse.pde.ui.tests_linux.gtk
org.eclipse.pde.ui.tests_winxp
org.eclipse.releng.tests_linux.gtk
org.eclipse.releng.tests_winxp
HTTP Download:
http://download.eclipse.org/downloads/drops/I20050412-1547
FTP Download:
user: anonymous
password: (e-mail address or leave blank)
server: download.eclipse.org
cd to directory: I20050412-1547
or
ftp://download.eclipse.org/I20050412-1547
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
