Menu
▾
▴
Re: [Iometer-devel] Case insensitive filesystems
|
From: Ming Z. <mi...@el...> - 2005-08-03 23:20:47
|
On Thu, 2005-08-04 at 00:49 +0200, Daniel Scheibli wrote: > Hi Ming, > > I also agree with that directory structure. So given > that no one speaks up against it till middle of the > month, I would propose to put this into live. thx. > > As I know that you once gave it a try - what would yes, i did. > (based on that experience) be the effort of adopting > all the source, header and makefiles? (I have a few > patches pending which should go in first but I would > be able to apply them as bunch in a consolidated > action) > i think this is part of a series of refactoring procedure. i suggest we to do this after you merge u patches and roll out DS3. so people have latest code to play with while I can do this. splitting them into folder like this is easy, 1 full day is enough base on my experience. the actual big work is to check all stuff in common and make sure they are absolutely necessary. then after that, i plan to abstract timing class and introducing HPET http://www.intel.com/hardwaredesign/hpetspec_1.pdf into linux. without the abstract timing class, adding hpet will make code messier, without hpet, the problem in current linux rdtsc issue can not be fixed. so we have to refactor the code before we introducing new features like that. > Daniel > > > > Ming Zhang said: > > Thanks. My main intentions are > > > > 1) separate the iometer & dynamo > > 2) clarify the interface between iometer and dynamo in common and thus > > it can be a contract later. i mean a clear interface there and then > > both iometer and dynamo can reuse them. > > 3) the code will be further to be divided in iometer&dynamo to have a > > common, per os and per arch specific sub layer. > > 4) common code style. > > > > basically, the code can only survival if we can define the contract > > between each components. otherwise it will become harder and harder to > > support new os/arch. coder come and leave, thus the interface become > > more and more unclear. none of us clearly know if adding something, > > will it break other stuff. then this make a big change become very > > hard. > > > > ming > > > > > > On Sun, 2005-07-31 at 14:26 -0500, Carlos Querol wrote: > >> I like it. It's about time the code was separated along more > >> functional lines. > >> It will also help when we develop more client interfaces and > >> standardize the network comm protocol. > >> > >> Carlos > >> > >> > >> Ming Zhang wrote: > >> > >> >i propose this dramatic change on iometer dir structure. > >> > > >> >[mingz@amd64 src]$ ls * > >> >Dynamo.dsp Iometer.dsw Makefile-Linux.i386 Makefile- > >> >Linux.ppc Makefile-Linux.xscale Makefile-Solaris.common > >> >Makefile- Solaris.sparc > >> >Iometer.dsp Makefile-Linux.common Makefile-Linux.ia64 Makefile- > >> >Linux.x86_64 Makefile-NetWare.i386 Makefile-Solaris.i386 TODO > >> > > >> >binary: > >> > > >> >common: > >> >IOAccess.cpp IOCommon.h IOMessage.h IOPort.h IOPortTCP.h > >> >IOTime.cpp IOVersion.h vipl.h > >> >IOAccess.h IOGlobals.cpp IOPort.cpp IOPortTCP.cpp IOTest.h > >> >IOTransfers.h ostream64.h > >> > > >> >dynamo: > >> >ByteOrder.cpp IOCQ.h IOManager.cpp > >> >IOManagerWin.cpp IOPerformanceNetware.cpp IOTargetTCP.cpp IOVIPL.h > >> > Network.cpp VINic.cpp > >> >Dynamo IOCQVI.cpp IOManager.h iomtr_kstat > >> >IOTarget.cpp IOTargetTCP.h NetTCP.cpp Network.h > >> >VINic.h > >> >IOCompletionQ.cpp IOCQVI.h IOManagerLinux.cpp > >> >IOPerformance.cpp IOTargetDisk.cpp IOTargetVI.cpp NetTCP.h > >> > Pulsar.cpp IOCQAIO.cpp IOGrunt.cpp IOManagerNetware.cpp > >> >IOPerformance.h IOTargetDisk.h IOTargetVI.h NetVI.cpp > >> > Pulsar.rc IOCQAIO.h IOGrunt.h IOManagerSolaris.cpp > >> >IOPerformanceLinux.cpp IOTarget.h IOVIPL.cpp > >> >NetVI.h PulsarRC.h > >> > > >> >iometer: > >> >AccessDialog.cpp GalileoCmdLine.cpp GalileoView.h > >> >LegalBox.cpp ManagerMap.cpp PageDisplay.cpp resource.h > >> >Worker.h AccessDialog.h GalileoCmdLine.h ICF_ifstream.cpp > >> >LegalBox.h ManagerMap.h PageDisplay.h StdAfx.cpp > >> >WorkerView.cpp AccessSpecList.cpp GalileoDefs.h ICF_ifstream.h > >> > MainFrm.cpp MeterCtrl.cpp PageNetwork.cpp StdAfx.h > >> > WorkerView.h AccessSpecList.h GalileoDoc.cpp > >> >ICFOpenDialog.cpp MainFrm.h MeterCtrl.h PageNetwork.h > >> >TextDisplay.cpp > >> >BigMeter.cpp GalileoDoc.h ICFOpenDialog.h > >> >Manager.cpp PageAccess.cpp PageSetup.cpp TextDisplay.h > >> >BigMeter.h GalileoGlobals.cpp ICFSaveDialog.cpp Manager.h > >> >PageAccess.h PageSetup.h WaitingForManagers.cpp > >> >GalileoApp.cpp Galileo.rc ICFSaveDialog.h > >> >ManagerList.cpp PageDisk.cpp PageTarget.h > >> >WaitingForManagers.h GalileoApp.h GalileoView.cpp Iometer > >> > ManagerList.h PageDisk.h res Worker.cpp > >> > > >> >scripts: > >> >ccntmknod ccntmknod.README iom-indent > >> > > >> > > >> >On Fri, 2005-07-29 at 10:19 -0700, Rick Altherr wrote: > >> > > >> > > >> >>I noticed this a while back, but had forgotten about it. The > >> >>IOmeter source tree has a folder named src/Dynamo. On a > >> >>case-insensitive filesystem (such as HFS+), when the dynamo > >> >>executable is to be created, the compiler fails because it cannot > >> >>overwrite a directory. I've simply moved src/Dynamo aside on my > >> >>system (since it only contains VC++ files) and everything builds > >> >>fine. > >> >> > >> >>We need to either move that directory or change the output name/ > >> >>location of the dynamo binary. > >> >>-- > >> >>Rick Altherr > >> >>Architecture and Performance Group > >> >>408-974-2607 > >> >>ral...@ap... > >> >> > >> >> > >> >> > >> >> > >> >>------------------------------------------------------- > >> >>SF.Net email is sponsored by: Discover Easy Linux Migration > >> >>Strategies from IBM. Find simple to follow Roadmaps, straightforward > >> >>articles, informative Webcasts and more! Get everything you need to > >> >>get up to speed, fast. > >> >>http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click > >> >>_______________________________________________ > >> >>Iometer-devel mailing list > >> >>Iom...@li... > >> >>https://lists.sourceforge.net/lists/listinfo/iometer-devel > >> >> > >> >> > >> > > >> > > >> > > >> >------------------------------------------------------- > >> >SF.Net email is sponsored by: Discover Easy Linux Migration > >> >Strategies from IBM. Find simple to follow Roadmaps, straightforward > >> >articles, informative Webcasts and more! Get everything you need to > >> >get up to speed, fast. > >> >http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click > >> >_______________________________________________ > >> >Iometer-devel mailing list > >> >Iom...@li... > >> >https://lists.sourceforge.net/lists/listinfo/iometer-devel > >> > > >> > > >> > > >> > >> > >> > >> ------------------------------------------------------- > >> SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > >> from IBM. Find simple to follow Roadmaps, straightforward articles, > >> informative Webcasts and more! Get everything you need to get up to > >> speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click > >> _______________________________________________ > >> Iometer-devel mailing list > >> Iom...@li... > >> https://lists.sourceforge.net/lists/listinfo/iometer-devel > > > > > > > > ------------------------------------------------------- > > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > > from IBM. Find simple to follow Roadmaps, straightforward articles, > > informative Webcasts and more! Get everything you need to get up to > > speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click > > _______________________________________________ > > Iometer-devel mailing list > > Iom...@li... > > https://lists.sourceforge.net/lists/listinfo/iometer-devel > > > |