Difference between revisions of "Performance/Meetings/2009 12"

Latest revision as of 23:11, 17 March 2010

Performance/Meetings/2009/12/11

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

Date: 2009/12/11
Time: 16:26– 17:26
Meeting No.:

Agenda:

(4:32:51 PM) liheng: Hi all, Good morning/afternoon
(4:33:42 PM) sb_: hi
(4:34:41 PM) cd_oo: hi all :-)
(4:35:20 PM) Matthias: hi all, sorry for being late
(4:35:21 PM) liheng: Hi,all
(4:35:43 PM) liangjun: :) Hi,all
(4:35:59 PM) yugq: hi all! sorry for being late.
(4:37:57 PM) liheng: Okay, we start to update our status,and discuss about new benchmark-system and helper-code
(4:39:40 PM) liheng: My status:I am working for documents of benchmark helper-code to clarify all usage of macro and functions
(4:41:05 PM) Dieter_: sorry for beeing late
(4:42:09 PM) liheng: Hi Dieter, we just began to update our status.
(4:42:22 PM) Dieter_: Hi Liheng :-)
(4:44:14 PM) ***Matthias has had some other tasks this week, including a deadlock issue (thanks sb_ :-) ), and thus made only small progress with 'store' refactoring.
(4:46:49 PM) yugq: My Status: Benchmark System is online again. But you will be get test log generated by the helper code first. So the first important thing is which module we can put the helper code and must enable the code for all as soon as possible.
(4:48:33 PM) cd_oo: My status: I am currently working on an important internal task. Fortunately the show stopper issue is fixed now. As I will be in vacation from Wednesday next week, there won't be any update from my side this year.
(4:48:34 PM) sb_: yugq: I would suggest the "tools" module, as it is rather low-level, but outside URE.
(4:49:28 PM) yugq: sb_: OK. So I will put it into tools next week. And try to switch the CWS to hg.
(4:49:41 PM) Matthias: yugq: I agree with sb_ that 'tools' would just be fine for now (to start with)
(4:49:54 PM) liangjun: My Status: update two wiki http://wiki.services.openoffice.org/wiki/Performance/load_performance_analysis http://wiki.services.openoffice.org/wiki/Performance/load_performance_implement
(4:50:13 PM) yugq: sb, mhu: Thanks for the advice.:)
(4:50:22 PM) liangjun: and continue to debug odf load performance in single thread.
(4:51:43 PM) liheng: sb_:Yes, it is great for all diagram works. Do you think which problems you refered last week must been fixed when we commit helper code in tools module?
(4:53:34 PM) Matthias: liheng: I think it would be best to follow most of stefans advice, it should not be too much work to apply the suggest changes, I think, and would make the code much cleaner and easier to understand.
(4:54:07 PM) sb_: liheng: In tools, the code can more easily evolve over time, so it is no longer /that/ important that it is perfect (e.g., might contain as-yet unused code).
(4:57:06 PM) Matthias: liangjun: your two wiki pages are too long to read in the background now, can we get more time before we comment, please ?
(4:57:55 PM) liangjun: mhu: ok
(4:58:01 PM) liheng: Matthias,sb_: Some macro reserved for small benchmark test and improvment of diagram(like macro BENCHMARK_USERDEF)...
(4:59:27 PM) Matthias: liangjun: thanks (I'm trying to read in the background, but want to follow our discussion also :-) )
(5:01:11 PM) liangjun: mhu::P Please Never mind that my English.
(5:01:34 PM) liheng: so, I will document how to use the benchmark helper code to trace new case and view data on our website.
(5:01:49 PM) Matthias: liangjun: don't worry, your english is not so bad in fact
(5:02:06 PM) liangjun: mhu: thank you.
(5:08:48 PM) liheng: sb_:I not sure about the problem:"- getLocalMutex uses the broken double checked locking pattern. Use functionality from rtl/instance instead",can you give me some idea?
(5:11:05 PM) sb_: liheng: Depending on memory architecture, the pattern used to create the mutex just once can fail (mostly an academic problem these days, though). Therefore, there is code (with examples how to use it) in sal/instance.hxx to "do it right." I can post a code snippet how the code should look like best.
(5:18:00 PM) liheng: I jusk took a look on it, I see, we will change to this safe mode :)
(5:18:47 PM) sb_: liheng: great, just drop me a mail anytime if you have any further questions
(5:20:43 PM) liheng: Thank you very much. :)
(5:21:23 PM) liangjun: mhu: If you have some thing, we can exchange mail lists.
(5:23:13 PM) liheng: Okay, no more for me today.
(5:24:04 PM) liangjun: me too.:)
(5:24:23 PM) Matthias: liangjun: okay, will read and then see
(5:24:37 PM) Matthias: otherwise I have no more for today, also
(5:24:56 PM) liangjun: mhu: :)
(5:25:10 PM) liheng: Okay, that's all.
(5:25:24 PM) Matthias: bye all, have a nice weekend
(5:25:32 PM) yugq: bye all.
(5:25:36 PM) Dieter_: bye!
(5:25:41 PM) liheng: Bye all,have a nice weekend
(5:25:44 PM) cd_oo: Bye all and have a nice weekend
(5:25:58 PM) liangjun: Bye all
(5:26:02 PM) sb_: bye

Performance/Meetings/2009/12/04

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

Date: 2009/12/04
Time: 16:12– 17:19
Meeting No.:

Agenda:

(4:31:34 PM) liheng: Hi,all
(4:31:40 PM) cd_oo: Hi all :-)
(4:31:49 PM) liheng: Good morning / afternoon!
(4:32:18 PM) sb_: hi all
(4:32:42 PM) yugq: hi all
(4:34:26 PM) liheng: sb_:Thank you for your review of perfbenchmark01.
(4:34:48 PM) Matthias: Hi all, sorry for being late
(4:35:00 PM) liangjun: hi all :P
(4:35:34 PM) liheng: Hi, Matthias, you are not missing any information
(4:35:35 PM) sb_: liheng: you are welcome; my biggest concern would be to move the new API out of URE for now; then it could evolve more easily over time (and eventually move back to URE when need be)
(4:36:52 PM) Matthias: Hi liheng, so you have not yet started ?
(4:37:55 PM) liheng: Matthias:Yes, it begin now ;P
(4:38:37 PM) liheng: sb_:We put all helper code into rtl for reason of rtl_log is in same module :)
(4:40:38 PM) sb_: liheng: I understand that. But I think here the goal of keeping URE API as small as possible should be considered more important.
(4:40:58 PM) liheng: So, if you think it don't need in sal/rtl, we can certainly move them into other module.
(4:41:22 PM) sb_: That would be great. :) Thanks.
(4:42:35 PM) liheng: It's my pleasure.
(4:42:59 PM) liheng: Okay, let's update our status.
(4:44:13 PM) liheng: yugq:Did you update website of benchmark-system?
(4:44:28 PM) liheng: yugq: Did you update website of benchmark-system?
(4:45:48 PM) yugq: liheng: not yet. I'm working on compare test data and generate report now. Closely finished. I think it will be next week to online;-)
(4:48:29 PM) liheng: My status:I did a little work on changed part saving and document for benchmark helper-code.
(4:49:06 PM) yugq: liheng: I did some change on the benchmark helper code this morning according sb_ suggestion. Not commit yet. We can talk about it next week.
(4:50:43 PM) cd_oo: My status: Nothing new from my side. I am currently working on OOo 3.2 show stoppers. Hopefully next week I have more time.
(4:51:41 PM) liheng: Matthias:For work of "saving" I were blocked by Calc because cell-referencing and formula, I think I will send a set of patch for text and draw to you first.
(4:52:21 PM) Matthias: okay, will have a look at it
(4:53:19 PM) liangjun: My status: I 'm working on the odf load performance still. and I'd finished almost impress code. testing and debugging.
(4:53:21 PM) liheng: yugq:I saw the e-mail today, and I reply it about performance_diagram.h and header generation tool
(4:53:32 PM) ***Matthias has had a day of vacation this week, and many other things to do; so not too much progress with store refactoring
(4:55:01 PM) sb_: I haven't got anything to share, either. CWS sb111 is in QA, where they keep finding minor issues...
(4:55:11 PM) liheng: yugq:I think we must work out the manual for benchmark-system and helper code as soon as possible :)
(4:55:36 PM) liangjun: And I'm know the sd and sfx2 little, It is hard for me
(4:56:14 PM) yugq: liheng: OK. I see.
(4:56:58 PM) liangjun: liheng: and the benchmark introduce :P
(4:57:22 PM) liheng: ALL: I have a question. Do we need to submit some issues of our work?
(4:59:06 PM) liangjun: liheng: Yes, so that I can submit my now source code to cws.
(5:00:33 PM) Matthias: yes, you should probably submit some or at least one issue to document your work
(5:02:22 PM) liheng: Matthias: Yes, it is exactly what I means.
(5:03:11 PM) liheng: I think we can trace our work with issuezilla.
(5:06:45 PM) liheng: No more topic for me, today.
(5:06:53 PM) yugq: All: From the test result shown on http://wiki.services.openoffice.org/wiki/Performance/Report/Perfomrance_Report_for_Startup , I think we need sum up the work did on from OOo3.1 to OOo3.2 ...
(5:07:03 PM) yugq: All: thus we can see how performance influenced by which effort. And also accumulate experience on maitaining OOo performance later.
(5:08:24 PM) yugq: liheng: sorry for not see the time. We can talk about that next IRC.
(5:09:08 PM) yugq: :)
(5:10:46 PM) liheng: yugq:No, I only mean no more for me,but if you have any topic, please! :)
(5:11:40 PM) liangjun: Where are I can learn the issu and cws . :-(
(5:13:02 PM) liheng: liangjun:I will get some information from some people of old Open Source Dept. in our company. (5:14:42 PM) yugq: liheng: I suggest we can set the topic for "sum up works" next week's IRC.
(5:14:43 PM) liheng: yugq:I think we can keep to discuss your topic in mailinglist.
(5:14:49 PM) liheng: ;)
(5:15:03 PM) yugq: liheng: yes.
(5:15:13 PM) liangjun: liheng: thanks, before yugq did that :P.
(5:15:15 PM) yugq: No more for me now.
(5:17:03 PM) liangjun: so me
(5:18:13 PM) Matthias: I dont have more for today, also
(5:18:59 PM) yugq: bye all
(5:19:02 PM) liheng: I think we can close this meeting early :)

Go back