Performance/Meetings/2007 12

From Apache OpenOffice Wiki
< Performance‎ | Meetings
Revision as of 23:13, 17 March 2010 by Ivanm (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
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: 2007/12/13
Time: 16:28– 17:47
Meeting No.:


(16:28:15) LiHeng: hello!:)
(16:33:00) xiuzhi: hi all
(16:33:14) mhu: Hi all, sorry for being late.
(16:33:29) LiHeng: mhu:hi mhu
(16:33:53) xiuzhi: hi all
(16:33:55) yugq: hello, everyone!
(16:34:08) LiHeng: mhu:did you receive the agenda for today
(16:34:26) mhu: yes, I've seen the agenda. Fine for me.
(16:35:12) LiHeng: are there some tipoc you want to insert to it?
(16:35:23) LiHeng: mhu:are there some tipoc you want to insert to it?
(16:35:38) mhu: No, I'm fine with the agenda as is.
(16:35:54) LiHeng: ok, let's start
(16:36:04) mhu: yes
(16:36:20) LiHeng: .Review the mission of Project discuss in previous IRCs
(16:37:00) LiHeng: for RF2000,the performance project will maintain a good performance of openoffice
(16:37:11) liangjun: hello .:)
(16:37:28) mhu: yes, agreed: maintain (longterm) good performancd
(16:38:06) mhu: s/performancd/performance/
(16:38:30) LiHeng: mhu::)it'll be mission for Preformance project
(16:38:54) LiHeng: xiuzhi:how do you think about it?
(16:39:42) xiuzhi: LiHeng:agreed
(16:40:10) LiHeng: main case for the missiom is build and maintain monitor,profiling tools and design test case to evaluate performance of OO
(16:40:43) mhu: yes
(16:42:03) LiHeng: ok, the last,Malte developed and introduced some performance monitoring framework for child workspaces here in hamburg, we will take them to OO.o or Beijing RF2000? (16:43:11) xiuzhi: mhu: where is malte?
(16:43:25) mhu: What I heard from Malte, is that he will first give the framework to you, only. No time to put it on OOo.
(16:44:10) mhu: xiuzhi: Malte and MBA said, they will be available via email, and join IRC only when we need them.
(16:44:36) mhu: i.e., they have too many meetings
(16:44:46) xiuzhi: mhu:ok
(16:44:49) LiHeng: mhu:ok, but we hope to get the code of those tools
(16:45:03) LiHeng: can we do it?
(16:45:13) mhu: LiHeng: yes, that is what Malte said to me. Don't know when.
(16:45:46) mhu: I will ask Malte for more details.
(16:45:59) LiHeng: mhu: i will send a mail to Malte ,too
(16:46:24) mhu: LiHeng: okay, that will help.
(16:46:36) LiHeng: that's all to tipoc 1
(16:46:54) LiHeng: Are we going to the next topic?
(16:46:59) mhu: okay, I think the mission is okay
(16:47:21) LiHeng: Project work scope
(16:47:43) LiHeng: 2.1 Maintain tools for performance 2.2 Evaluating OpenOffice 2.3 Analyze report from end-user and testing 2.4 Work out proposal for optimizing and refactoring
(16:47:45) xiuzhi: mhu:okey
(16:48:57) LiHeng: mhu: do you think these working can make OO in a good performance status? (16:49:31) yugq: I tried VTune on Windows and Valgrind on Linux last week, they worked well and I think thease tools can help us.
(16:49:36) mhu: LiHeng: can I write a longer reply? give me a minute pease.
(16:49:55) mhu: s/pease/please/
(16:50:17) LiHeng: yes , i'm ok
(16:51:07) mhu: 2.1 tools: I usually select the tools that best answer my question, i.e. I first have the question than select the tools...
(16:51:34) mhu: ...this is because different tools can help you with different questions...
(16:51:48) yugq: mhu: yes
(16:52:32) mhu: ...e.g. valgrind/cachegrind/callgrind /Vtune help you when you want to know the callstacks for a particular problem...
(16:52:35) LiHeng: mhu:yes, we can list tools for all developers,and analyze for very major vesion
(16:52:59) mhu: but they generate too much data, when you do not yet know where to look.
(16:53:04) yugq: mhu: valgrind has many tools to do different analization in different aspects of performance.
(16:53:51) mhu: okay, I see that you already have worked with these tools and know what they can do and when they might help.
(16:54:32) yugq: mhu: do you think the first thing we need do is knowing the problem and then do profiling?
(16:54:55) mhu: yugq: yes, that is what I'm trying to say.
(16:55:20) LiHeng: mhu:okey
(16:55:39) yugq: mhu: ok
(16:56:06) mhu: That's also the reason, why I always asked for your "end-user problem" spreadsheet.
(16:56:31) LiHeng: yes , next Analyze report from end-user and testing
(16:56:54) mhu: yes
(16:57:56) LiHeng: Project would make the report to analyze the problem from end-user
(16:58:28) mhu: I think we know the two major effects (startup + doc. load/save times), but do we know what is worst?
(16:59:00) mhu: LiHeng: yes, we can start with that report to analyze problem.
(16:59:41) LiHeng: we would collect problem again in beijing
(16:59:45) xiuzhi: mhu: IMHO load/save is worst
(17:00:24) mhu: xiuzhi: yes, that is also what I read from your spreadsheet.
(17:00:26) xiuzhi: mhu: sorry. I misunstand you
(17:00:37) LiHeng: xiuzhi: startup load/save are easy to look, but maybe not be worst?:)
(17:00:48) yugq: I think it's hard to say what is worse to end-user. Both worst.;-)
(17:01:33) xiuzhi: mhu: What is "what is worst" mean?
(17:01:48) mhu: yugq: maybe it is the combination: double click doc on desktop; application starts; loads document.
(17:02:18) yugq: xiuzhi: maybe "what is the worst thing"
(17:02:46) mhu: xiuzhi: means: what does a enduser see as the worst problem, and what do we analyze to be the largest effect.
(17:03:00) yugq: mhu: yes
(17:03:02) xiuzhi: mhu: ok
(17:03:18) LiHeng: would we be going to topic 3, and in roadmap of project we can set the target for the analysis of profiling and problems?
(17:03:50) mhu: means: enduser sees long doc load time, but we analyse app starts too slow, but then loads doc fast (just scenario)
(17:04:38) xiuzhi: mhu: I see now
(17:04:44) yugq: mhu: yes
(17:04:58) mhu: okay, I cant typw / write so fast :-(
(17:05:10) mhu: s/typw/type/
(17:05:28) mhu: I'm ready for next topic, now.
(17:05:45) LiHeng: mhu: yes, we must differ the feel and the fact!!
(17:06:21) LiHeng: xiuzhi:can we be going to next ?
(17:06:38) mhu: LiHeng: yes; you see, I have been a scientist in my former life :-)
(17:06:53) xiuzhi: LiHeng: yee
(17:07:24) mhu: LiHeng: yes, next topic
(17:07:27) LiHeng: 3.Roadmap/Schedule & Routine
(17:08:37) mhu: I leave the Roadmap to you; I will help as much as I can, but you do the work. So you say what you want.
(17:08:45) xiuzhi: LiHeng: do you have any proposal about the roadmap?
(17:10:10) LiHeng: mhu: first some Routine, we want make a report to performance for every major version of openoffice,
(17:10:43) mhu: yes, such a report will help to analyze.
(17:12:22) mhu: but instead of major version, what do you think of the quarterly releases, i.e. 2.3.1, 2.3.0, 2.2.1, ...?
(17:12:23) xiuzhi: LiHeng: yes, it is the base of further analyzing
(17:12:26) LiHeng: and Project will set a series of number to explain the report and differ different version
(17:13:27) mhu: yes, we will have a series of numbers helping to explain what we see.
(17:13:28) xiuzhi: LiHeng: every release version is OK
(17:13:30) LiHeng: mhu:the dev version , i think
(17:14:08) LiHeng: we can make release version well
(17:14:24) mhu: LiHeng: okay, if we can do more versions (dev builds also) that will be fine. But this may not be necessary to start .
(17:14:41) xiuzhi: LiHeng: dev version is too many ,I think
(17:15:11) LiHeng: xiuzhi & mhu : we can select some stable version?
(17:15:16) yugq: I think we can focus on main versions first
(17:15:16) mhu: dev builds may be long term goal of "monitoring performance"
(17:16:15) mhu: LiHeng: yes, stable (public) releases like 2.3.1 and so on, will be fine for a start.
(17:16:18) LiHeng: mhu: aha, i see , i agree to release version , we can do better to next release:)
(17:16:28) xiuzhi: LiHeng: There willbe one to two stable dev version,
(17:16:54) xiuzhi: LiHeng: per month
(17:17:22) xiuzhi: mhu: agreed, That would be fine for us
(17:18:30) LiHeng: mhu: do you think our first step can start from 2.3.1?
(17:18:56) mhu: LiHeng: yes, that would be a good start, I think.
(17:19:08) xiuzhi: LiHeng: OOo 2.4 cannot be available until March. 2008
(17:19:41) mhu: yes, for 2.4 there are currently only weekly dev builds.
(17:20:25) LiHeng: mhu:ok , we make a schedule for first step now
(17:20:29) LiHeng: ;)
(17:20:33) xiuzhi: 2.3.1 Is a good choice if we will launch this project now
(17:20:47) mhu: perhaps we can also look at a second version, and try to understand what has changed, if e.g. startup is different.
(17:21:13) LiHeng: mhu:good point!:)
(17:21:39) yugq: yes, it will be a reference
(17:21:45) LiHeng: mhu: please choice the second for us
(17:22:15) mhu: LiHeng: maybe, I can find some old data to show the history of OOo 1.1 -> 2.0 (17:22:32) mhu: LiHeng: yes, I will think about a second version.
(17:22:46) LiHeng: thank you
(17:22:48) yugq: mhu: it will help us much!
(17:23:37) LiHeng: mhu:we will make the report for 2.3.1 in two mouth,
(17:24:03) LiHeng: and analyze it to show worst
(17:24:30) mhu: okay, fine.
(17:24:48) LiHeng: then combine the end-user problem to the report
(17:25:19) LiHeng: sorry , in two months
(17:25:44) mhu: LiHeng: yes, understood; 2 months
(17:26:32) LiHeng: in this period , we will also do documentation for RTL and other profiling tools
(17:27:34) mhu: yes, that will also help; I think, Malte and I have never document our experience with the tools
(17:27:52) mhu: s/document/documented/
(17:27:54) LiHeng: for the details of the schedule, i will give you a mail, and some cases we need Malte's help
(17:28:35) mhu: okay; yes, Malte (and others) will help when needed.
(17:28:52) LiHeng: yes ,yes ,yes
(17:29:03) LiHeng: going to last topic?
(17:29:08) mhu: okay
(17:29:22) LiHeng: a little topic: Setup Mailinglist and Homepage of Project
(17:30:18) LiHeng: mhu: do we need the mailinglist and homepage on openoffice now?
(17:30:32) xiuzhi: mhu: ,meeting minutes and documents involved in should be public on wiki page . and the mailing list and so on
(17:30:49) xiuzhi: mhu: who is responsible for this?
(17:30:54) mhu: I know, you already work with other teams, e.g. the framework (mba, cd) team. How do you communicate there?...
(17:31:30) mhu: xiuzhi: yes, wiki as webpage and for minutes is good (and preferred choice)
(17:32:00) LiHeng: mhu: who can we contact to?
(17:32:01) xiuzhi: mhu: there is a @framework mailing list and @dev..
(17:32:27) xiuzhi: mhu: we need a independent mailling list for this project
(17:32:53) mhu: xiuzhi: okay, dev@framework is what I thought. ..
(17:33:00) xiuzhi: mhu: we will use dev@... and @ performance ... That would be fine
(17:33:03) xiuzhi: mhu:yes
(17:33:28) mhu: do we really need a different mailing list? I think, the framework team can be of good help anyway.
(17:34:17) xiuzhi: mhu:Yes. we need because the performance is an independent project.
(17:34:34) mhu: xiuzhi: okay, I see your point...
(17:35:01) mhu: ...let me think how we can have that independent mailing list...
(17:35:33) mhu: ...current ooo infrastructure cannot easily create such a separate mailing list...
(17:35:33) xiuzhi: mhu: new layout manager is included in framework , so it is not necessory for that project, that is not really a project
(17:36:15) mhu: ...let me talk to Martin and Stefan Taxhet how we can create some mailing list for this project.
(17:36:29) LiHeng: xiuzhi:maybe we can create a wiki and homepage only?
(17:36:36) xiuzhi: mhu: okey, thanks a lot
(17:36:51) xiuzhi: LiHeng: not enough for us
(17:37:12) xiuzhi: LiHeng: we need more help from all over the OOo community
(17:37:15) mhu: LiHeng: we can start with the wiki now / today / next week. a mailing will take some time.
(17:37:28) xiuzhi: LiHeng: so mailinglist is necessory
(17:37:45) xiuzhi: mhu:okey, a good idea
(17:38:02) LiHeng: mhu:thanks a lot
(17:38:05) xiuzhi: LiHeng: so we can create a wiki page for this project now
(17:38:33) LiHeng: xiuzhi:do you know how can make it?
(17:38:41) xiuzhi: LiHeng: yes
(17:38:57) LiHeng: everything is ok
(17:39:14) xiuzhi: LiHeng: Peter and Yunan can do a help, you can ask them tomorrow
(17:39:34) LiHeng: mhu: we make a big step ;)
(17:39:45) LiHeng: thank you!
(17:40:00) mhu: LiHeng: yes, also thank you!
(17:40:15) LiHeng: mhu:so, that's all today for me
(17:40:21) mhu: okay, I have one last topic for today...
(17:40:42) LiHeng: ok,please!
(17:40:58) mhu: ...I am out of office for the next 3 weeks (holidays for xmas and new year)...
(17:41:17) mhu: I wont be available until 2nd week of January.
(17:41:40) yugq: mhu: Wish you have a good holiday!:-D
(17:42:12) LiHeng: mhu:from next week to 2nd of jan?
(17:42:24) mhu: yes. so, the next meeting for me would be Thursday, ~10th January (don't have the exact date).
(17:42:34) liangjun: mhu:ya , Wish good holiday.
(17:42:38) xiuzhi: mhu: ok, have a good trip
(17:42:54) LiHeng: mhu:best wish!
(17:43:17) mhu: okay, thank you all for your good wishes. I wish you all the best also. See you next year.
(17:43:58) xiuzhi: mhu: we can use this peoried to prepare the document and setup the wiki page and start analyzing...
(17:44:10) xiuzhi: mhu: see you
(17:44:11) LiHeng: mhu:we have a request
(17:44:23) mhu: LiHeng: yes?
(17:45:03) LiHeng: can you send us a photo of you, let us know you farther?
(17:45:16) xiuzhi: LiHeng: I have one
(17:45:34) xiuzhi: LiHeng: but not him only.
(17:45:55) mhu: LiHeng: okay, let me see if I can find one. But I think Peter Junge may have a photo of mine. You can also ask him.
(17:46:12) LiHeng: okey, that's all
(17:46:41) mhu: okay, and good bye.
(17:46:55) LiHeng: bye,
(17:46:57) yugq: bye!
(17:46:59) xiuzhi: bye
(17:47:06) liangjun: bye :)

Go back

Personal tools