Performance/Meetings/2009 11

From Apache OpenOffice Wiki
Jump to: navigation, search

Performance 170.png
Performance Project

Quick Navigation




About this template


Meeting Minutes
IRC Meeting of Sun Microsystems (StarOffice) with RedFlag2000
Performance Project

Date: 2009/11/27
Time: 15:38– 17:50
Meeting No.:


(4:59:21 PM) Matthias: Hi all
(4:59:30 PM) cd_oo: hi all :-)
(4:59:50 PM) liheng: hi,all
(5:00:02 PM) Dieter_: Hi all :-)
(5:02:00 PM) yugq: hi all
(5:07:04 PM) yugq: my status: the load/save test point are inserted into the new CWS perfbenchmarker01. I'm working on the benchmark system for now, and it will be online after weeks.
(5:07:52 PM) sb__: yugq: I saw that you created that CWS, and have some comments on it:
(5:08:23 PM) sb__: First, I think it is not a good idea to still create an SVN CWS, instead of a HG one.
(5:09:45 PM) liheng: Dieter_:First, I want to comfirm the time of this meeting, 8:00am UTC is 16:00am in Beijing,but what's time in hamburger
(5:10:22 PM) liheng: and are you in the winter time?
(5:10:28 PM) yugq: sb__: Sorry, I didn't notice the CVS system change. I will check it out. Thanks for the information.
(5:10:29 PM) sb__: Second, I always get nervous ;) when things are added to the sal API. Since we have to keep that API stable, we have to be very careful with what we add there. I did not look at your files in detail yet; is there a rationale to add the things to sal (instead of some other module further up)?
(5:11:27 PM) Dieter_: Liheng: It's 10am Hamburg time
(5:11:35 PM) sb__: Third, I think there are some things that need to be cleaned up, like old LGPL2 headers (or no headers at all); .h files that use C++ inline constructs without #ifdef __cplusplus around them, things like that.
(5:11:51 PM) Dieter_: Liheng: yes, we have winter time
(5:12:42 PM) yugq: sb__: Our code in sal is used to log time which quite like RTL_LOG, I think it will not influence other APIs.
(5:14:07 PM) yugq: sb__: code in sal is written by liheng. liheng: how about your ideas?
(5:16:03 PM) yugq: sb__: I think we can improve the code as soon as possible. Thank you for your check on it:)
(5:16:20 PM) liheng: Dieter_:Can we shift to an earlier time, for 30mins to this meeting? Because we will off from work in 17:30,Beijing time, some people must take work bus
(5:16:40 PM) sb__: yugq: I can try to get some time next week to look at it, and send you further comments.
(5:18:00 PM) Dieter_: Liheng: Yes, that would work for me personally but what's about the others? Opinions?
(5:18:04 PM) yugq: sb__: That will be wonderful!
(5:18:10 PM) liheng: Matthias:"perfbenchmarker01" is for BenchmarkSystem,in order to record benchmark information.
(5:18:39 PM) sb__: Dieter_, Liheng: 9:30 Hamburg time would be fine with me.
(5:20:08 PM) cd_oo: 9:30 hamburg time is fine for me, too
(5:21:07 PM) Matthias: 9:30 si okay for me also
(5:21:14 PM) Matthias: s/si/is/
(5:22:03 PM) Matthias: liheng: yes, I talked to sb yesterday about perfbenchmarker01, I will have a look also some time next week
(5:24:04 PM) liheng: My status:I am sorting out code for parted saving /loading rebase to recently codebase.
(5:26:18 PM) cd_oo: My status: I am currently working on OOo 3.2 show stoppers therefore no time to do something for performance.
(5:27:01 PM) liheng: Yugq and liangjun must leave for work bus :)
(5:28:03 PM) Dieter_: Bye!
(5:28:18 PM) sb__: My status is that the configmgr rewrite is in QA currently, and I plan to make builds of it available, probably next week, so that people can try it out.
(5:29:42 PM) ***Matthias looked at a "buffered file I/O" show stopper this week; otherwise module "store" refactoring for smaller block (and thus file) sizes.
(5:33:42 PM) Matthias: liheng: as it is getting late for you in Beijing, shall we finish for today (and start 1/2 hour earlier next week) ?
(5:33:43 PM) liheng: What's the work of "store" refactoring?
(5:35:04 PM) liheng: I don't need to take bus,so I can continue the meeting, and we start 1/2 hour earlier next week :)
(5:35:37 PM) Matthias: liheng: module "store" manages the *rdb file files (types.rdb, services.rdb), is fixed block size based, and thus file size determined by number of blocks times block size; with smaller block size, we can have smaller file size, and thus faster (cold) startup.
(5:37:53 PM) liheng: Matthias:I see, for small size of rdb we can load it faster
(5:38:00 PM) Matthias: yes
(5:39:13 PM) Matthias: current estimate is new file size could be 1/2 to 1/3 of current file size, but we'll see
(5:41:52 PM) Matthias: I don't have more for today ...
(5:42:13 PM) liheng: All performance report in my hand show that we can't get time form QuickStarter of OOo, It's only save 10% time from cold-start with Quickstarter ...
(5:43:14 PM) Matthias: can you send email with some data, and also describe how you measured, please ?
(5:43:37 PM) liheng: I review the changes of OOo I guess with new URE Quickstart only load a few libraries ?
(5:43:48 PM) Matthias: ...then others can look into that also
(5:45:41 PM) Matthias: ...maybe cd_oo has some time after taking care of 3.2 stopper bugs ?
(5:46:25 PM) cd_oo: yes, I can have a look at it after I fixed my show stoppers
(5:46:40 PM) liheng: Matthias: Yes, I send the section for Quickstarter next monday
(5:46:54 PM) Matthias: ...without more info, and looking into it, I could only guess what's going on with quick start.
(5:47:53 PM) Matthias: liheng: sending to the list would be fine, so cd_oo can have look also
(5:48:15 PM) cd_oo: yes, please send your findings to the performance list
(5:48:27 PM) liheng: Okay, no more for me,today
(5:49:11 PM) Matthias: okay, then see you next week, 1/2 hour earlier (I will fix our Hamburg calender entry accordingly)
(5:49:47 PM) liheng: Thanks, and bye
(5:50:00 PM) liheng: Bye all, nice weekend:)
(5:50:09 PM) Matthias: bye all, have a nice weekend
(5:50:19 PM) cd_oo: bye all and have a nice weekend

Go back

Personal tools