Vm-dev March 2006

vm-dev@lists.squeakfoundation.org

14 participants
19 discussions

[BUG]? Compiling on linux sparc64
by Damien Pollet 21 Mar '06

21 Mar '06

Hi, Last night I tried to build 3.8a-1on linux Ubuntu Dapper on an Sun Ultra5. I had to make a small change in plugins/SerialPlugin/sqUnixSerial.c to make the build work... not sure if it's a problem with my setup or an actual bug. There was nothing above B2000000 defined in <bits/termios.h>, so here is the change I made: --- platforms/unix/plugins/SerialPlugin/sqUnixSerial.c 2006-03-21 08:39:36.000000000 +0100 +++ /home/cdlm/tmp/Squeak-3.8a-1/platforms/unix/plugins/SerialPlugin/sqUnixSerial.c 2006-03-21 11:20:03.000000000 +0100 @@ -95,6 +95,8 @@ { 1152000, B1152000 }, { 1500000, B1500000 }, { 2000000, B2000000 }, +# endif +# if defined(B2500000) /* missing on Ubuntu Dapper, 2.6.15-19-sparc64 */ { 2500000, B2500000 }, { 3000000, B3000000 }, { 3500000, B3500000 }, -- Damien type less, do more

1 0

Fwd: Squeak on UTF-8 system
by John M McIntosh 20 Mar '06

20 Mar '06

Begin forwarded message: > From: Martin Kuball <martinkuball(a)web.de> > Date: March 20, 2006 10:38:03 AM PST > To: The general-purpose Squeak developers list <squeak- > dev(a)lists.squeakfoundation.org> > Subject: Re: Squeak on UTF-8 system > Reply-To: The general-purpose Squeak developers list <squeak- > dev(a)lists.squeakfoundation.org> > > Am Friday, 17. March 2006 19:51 schrieb Martin Kuball: >> Am Thursday, 16. March 2006 21:10 schrieb Petr Fischer: >>>> About 1 and a half year ago I submitted a patch for the unix >>>> keyboard input handler of the VM to make it utf8 ready. >>>> Unfortunately it never got included. Maybe it's time to try >>>> again. >>>> >>>> Martin >>> >>> Share it please. Thanks! pf >> >> I don't have much time this weekend. But I will try to dig up the >> stuff and post it here. > > I'm not sure if the patches are still needed. Some previous mails > suggested that it's possible to make it work by specifing the right > parameters on the comandline. Well, it does not work for me. So find > attached to this mail the patches I made over a year ago. > Unfortunately I wasn't able to test them with a recent VM. I tried > with the svn version but it failed to compile because of some missing > methods (like fetchLong32ofObject). > > Martin > -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

1 0

OpenGL in MacOS browser plugin
by Josh Gargus 19 Mar '06

19 Mar '06

Hi, After failing with the latest subversion sources, I was finally successful in compiling a MacOS browser plugin based on the VMM38b4 sources. However, I can't get OpenGL to work. I've traced the problem to sqMacOpenGL.c. The call to GetWindowGreatestAreaDevice() fails with an error code of -5600 ("Invalid Window Pointer"). I assume that this is because getSTWindow() doesn't work when Squeak is running in a browser plugin. I hacked it by avoiding all the GDH stuff if we are in a browser, like so: #ifdef BROWSERPLUGIN swAttrib[2] = 32; #else #ifdef INTERNAL #if TARGET_API_MAC_CARBON DPRINTF(5, (fp, "Getting GDH (Carbon): %d\n", getSTWindow(), flags)); stat = GetWindowGreatestAreaDevice(getSTWindow (),kWindowContentRgn,&tempGDH,&ignore); if (stat) DPRINTF(1, (fp, "ERROR: failed call to GetWindowGreatestAreaDevice (code: %d)\n", stat, flags)); #else DPRINTF(5, (fp, "Getting GDH (non-Carbon)\n", flags)); tempGDH = getDominateDevice(getSTWindow(),&ignore); #endif if (tempGDH == nil) { DPRINTF(1, (fp, "ERROR: tempGDH is nil (whatever that means)\n", flags)); return -1; } swAttrib[2] = (*(*tempGDH)->gdPMap)->pixelSize; #else swAttrib[2] = (*(*GetMainDevice())->gdPMap)->pixelSize; #endif if(swAttrib[2] < 16) swAttrib[2] = 16; #endif Obviously, this is crap, but I don't know the right thing to do. Anyway, this allows me to successfully create an OpenGL context ("OpenGL newIn: aRect"). Adding an external FFI plugin to the browser plugin directory lets me make OpenGL calls. But something is still wrong: it seems like the OpenGL buffers are never swapped (although DPRINTF assures me that they are: "### Swapping renderer buffers"). I've had enough of this for tonight; I need some sleep. But in the meantime, I'll take any advice I can get (eg: "stop working on that, the latest sources will do everything you need RSN"). Thanks, Josh

2 1

Mac Carbon VM goes to 3.8.11b4
by John M McIntosh 19 Mar '06

19 Mar '06

I've put up 3.8.11b4 universal to fix a problem reported by dual screen users. Cmd-F does not show the context menu in the right place. Also note I altered the mpeg3plugin to place 0xFF in the alpha channel to fix problems with 32 bit movies under Tweak. Please find the VM in the usual places via the link http://www.smalltalkconsulting.com/squeak.html 3.8.11b4 compiler should asks for inlined sqSqueakMemory accessors if intel, these generate better assembler than non-inline which is better for powerpc. sqMacMain.c Browser plugin code must return VM Path name, can't use argv[0]. sqMacMemory.c (see 3.8.11b3, perhaps not checked in correctly) sqMacNSPlugin.c Start poking at code to make work under browser again sqMacUIEventsUniversal.c old getmousepoints should not convert coord to global, on cmd-key remember old coordianates versus doing getmousepoints since the port is not set. Likely we've stored the last mouse move! mpeg3plugin audio.c output.c Change to set Alpha to OxFF versus 0x00 in order to play better with Tweak in 32bit mode misc poking at MMX (not working) for MacIntel -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

2 1

Pending carbon mac VM universal 3.8.11b3U
by John M McIntosh 18 Mar '06

18 Mar '06

After many many hours of looking at evil intel code and various internet source we discovered that Mac Solo CPUs are quite sensitive to instruction alignment issues. By tweaking the compiler a little, ignoring the obsolete gnu register overrides, plus providing march and tune commands more fitting to the Solo this results in a VM that is 20% faster for bytecodes and 15% faster for Sends than the earlier shipped 3.8.11b2U version. Beyond that there are also some minor changes: a) Use PageSize boundary for allocating start of Object memory heap (likely never used). b) Alter getSystemAttributes: 0 to return the full path to the executable, versus the path to the application. Look for the zip file Squeak 3.8.11beta3U.app.zip in the usual places via the link http://www.smalltalkconsulting.com/squeak.html PS we wait for the Intel compiler which has always promised much better compiled code, likely it seems this summer. -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

1 0

Croquet VM issues
by Andreas Raab 17 Mar '06

17 Mar '06

Hi Folks - Since we're nearing release status I was going over the VM requirements for Croquet to see if we could just go with a bunch of standard VMs or whether we'd need some customized ones. Looking over some recent VMMaker versions, it turns out that there are actually a few important bits and pieces missing from the current VMMaker release which we will want to use in Croquet. Most importantly, we need the following issues resolved: 1) GC improvements: As discussed in April last year (see http://discuss.squeakfoundation.org/cgi-bin/ezmlm-browse?list=vm-dev&cmd=sh…) we need the weak array fixes and the GC measurements/tweaks that John did. Those actually turn out to be important for us, if only for analysis of current behavior. Questions: * What's the general status of those changes? The way I understand the discussion we had back in april, is that John did the adjustments to 64bit and then ... what? * @Tim: Does a VMMaker repository exist by now? Like I was saying back then, I am not alltogether happy with the current situation of new releases just popping up on SqueakMap and with no way of seeing what's in there and how things have developed. 2) Floating point accuracy: As discussed in november last year (see http://discuss.squeakfoundation.org/cgi-bin/ezmlm-browse?list=vm-dev&cmd=sh…) we need consistent floating point behavior across platforms. I'm actually in the process or rolling out a bunch of tests for it which will be based on fdlibm as described in the above message. One issue that is critical here is that I am planning to checksum (SHA1) the VMs while we are running Croquet to see if the VM is "compliant" (if not we'd run a whole suite of tests to verify the behavior). Questions: * On Mac and Unix, do we implement getSystemAttribute: -1 to return the full path to the executable? * On Mac and Unix, can one read the executable while running Squeak? I need this to checksum the VM proper. 3) Some Socket fixes: David Reed made some changes for Sockets which turn out to be critical in particular on Macs and Unix systems. I'm still in the process of bundling up these changes but they are fairly straightforward so I don't see many issues with them. Besides the above three, I have one additional question: What is the general status of the 64bit work right now? Do we consider this complete or work in progress or experimental? Can we rule out any ill effects for the 32bit VMs? Thanks and Cheers, - Andreas

7 12

I'm unable to run squeak (Segmentation fault)
by Josef Hajas 10 Mar '06

10 Mar '06

Hello, I have IBM ThinkPad T42 (Pentium M), Gentoo with Linux kernel 2.6.14-gentoo-r2, using gcc 3.3.5-20050130. I have tried binary version, compile vm from sources (downloaded from http://www.squeak.org/Download/ - it is version 3.7.7) and finaly also install older version from portage. Always I get this: $ squeak Squeak3.8-6665full.image Segmentation fault -1796312376 FileDirectory>directoryContentsFor: -1796312480 FileDirectory>entries -1796312572 FileDirectory>fileNames -1796312948 FileDirectory>fileExists: -1796314856 FileDirectory class>openSources:forImage: -1796315684 FileDirectory class>openSources:forImage: -1796315776 FileDirectory class>openSources:andChanges:forImage: -1796317624 SmalltalkImage>openSourceFiles -1796326280 FileDirectory class>startUp -1796326436 Behavior>startUp: -1798257836 [] in SystemDictionary>send:toClassesNamedIn:with: -1798258060 OrderedCollection>do: -1798258152 SystemDictionary>send:toClassesNamedIn:with: -1798258244 SystemDictionary>processStartUpList: -1798446004 SmalltalkImage>snapshot:andQuit:embedded: -1798446096 SmalltalkImage>snapshot:andQuit: -1798446188 TheWorldMenu>saveAndQuit -1798446372 TheWorldMenu>doMenuItem:with: -1798446464 [] in MenuItemMorph>invokeWithEvent: -1798446556 BlockContext>ensure: -1798446648 Cursor>showWhile: -1798446868 MenuItemMorph>invokeWithEvent: -1798446964 MenuItemMorph>mouseUp: -1798447056 MenuItemMorph>handleMouseUp: -1798447320 MouseButtonEvent>sentTo: -1798447412 Morph>handleEvent: -1798447504 MorphicEventDispatcher>dispatchDefault:with: -1798447596 MorphicEventDispatcher>dispatchEvent:with: -1798447688 Morph>processEvent:using: -1798447780 MorphicEventDispatcher>dispatchDefault:with: -1798447888 MorphicEventDispatcher>dispatchEvent:with: -1798448020 Morph>processEvent:using: -1798448112 Morph>processEvent: -1798448204 MenuMorph>handleFocusEvent: -1798448480 [] in HandMorph>sendFocusEvent:to:clear: -1798448388 [] in PasteUpMorph>becomeActiveDuring: -1798448572 BlockContext>on:do: -1798448664 PasteUpMorph>becomeActiveDuring: -1798448848 HandMorph>sendFocusEvent:to:clear: -1798448940 HandMorph>sendEvent:focus:clear: -1798449084 HandMorph>sendMouseEvent: -1798449176 HandMorph>handleEvent: -1798449448 HandMorph>processEvents -1798449356 [] in WorldState>doOneCycleNowFor: -1798449540 SequenceableCollection>do: -1798449632 WorldState>handsDo: -1798449724 WorldState>doOneCycleNowFor: -1798449816 WorldState>doOneCycleFor: -1798449908 PasteUpMorph>doOneCycle -1798576280 [] in Project class>spawnNewProcess -1798576188 [] in BlockContext>newProcess Neúspěšně ukončen (SIGABRT) Before segmentation fault for very short moment window is displayed. I'm attaching strace. Any idea how to make seqak working on my computer? Thank you Josef Hajas

1 0

Mac Carbon Universal VM for MacIntel
by John M McIntosh 08 Mar '06

08 Mar '06

Ok, first I have to thank Karsten Wolf for running strange and interesting VMs on his Mac Intel to get us started on the conversion process. Second I'd like to thank Impara GmbH for not only producing Plopp (http://www.impara.de/project_painter.htm) with Squeak, but also for giving me a Mac Mini Core Solo to help in the conversion process, because of that machine and a sleepless night or two, we now understand why we have the 16 bit screen depth conversion issue, and why full screen did not work. Later tonight after some rest, reflection and a bit more testing I'll post a 3.8.11b2 Mac Carbon Universal Squeak VM for regular distribution as we are seeing more macintel squeak users. As earlier mentioned I still have some plugins to recompile, so the effort will be ongoing for a few weeks. 1 tinyBenchmarks shows on my old PowerBook 17inch 1.5Ghz. '113,274,336 bytecodes/sec; 5,111,787 sends/sec' On the Mac Mini Solo (also 1.5Ghz) that VM running under Rosetta does ' 43,301,759 bytecodes/sec; 2,056,941 sends/sec' As you can see the performance is 1/2 speed of the powerpc, not bad for emulation, but ugly for general usage. A universal VM does '226,548,672 bytecodes/sec; 5,811,339 sends/sec' and Karsten reports the graphics performance on his Core Duo is stellar. On 7-Mar-06, at 2:45 PM, Karsten Wolf wrote: > Hi, > > > Am 07.03.2006 um 20:17 schrieb Mike: > >> It seems to be slow on mine anyway. Windows take a long time to >> appear for example, so maybe its morphic thats slow. Or maybe its >> the entire VM. Any one else experiencing the same? > > No. > > Even in rosetta emulation the intelmac vm was at least as fast as > my powerbook g4. > > Try a different image. > > If you want to try an intel binary, see: > > http://groups.yahoo.com/group/squeak/message/103433 > > Make sure that the image you use is set to: > -32 bit colors; set monitor to 32 bit for best performance > -fullscreen off; the intelmac VMs can't do fullscreen > > -karsten > > -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

1 0

mac carbon vm 3.8.11b1U posted to experimental
by John M McIntosh 07 Mar '06

07 Mar '06

I've posted a mac carbon vm 3.8.11b1U for universal to http:://www.smalltalkconsulting.com/squeak.html follow links to idisk or FTP. This VM is a result of a major cleaning done over the weekend where I did: a) Refactored removed old OS-9 code, always use carbon, etc, basic housekeeping chores that were put off for a few years. b) Removed TARGET_API_MAC_CARBON, IHAVENOHEAD c) Made most variables static, procedures static where applicable. d) Always use Unix file names, drop any HFS logic e) Eradicate most usages of FSSpec, except for drop file logic (sigh) and a specialized use in InternetConfig, and URL post/get f) Reworked much of drag and drop logic to eradicate complaints from the 10.4u depreciated warnings monster.... Testers should ensure drag and drop works as expected from the finder, and Safari, also that starting the VM by cmd line, by clicking on one or more images, or squeak documents, or drag and drop to squeak VM in finder works as expected, and or setting the plist.info Also someone might confirm that the MacIntel box likes this VM, untested here on the island. I'll note I did poke a bit at the full screen logic for the MacIntel port, perhaps it works now? I have a NPSqueak (the squeak browser plugin) but need some brave fellow to poke at it since it's been converted to unix file names and refactored and greatly touched and untested in the area of URL GET and POST logic, let alone a zillion other areas. Likely as time permits I am going to do a paranoid pass thru the Mac VM code to ensure bad return codes, improper values are considered, trapped and give acceptable feedback and fprintfs to stderr. For the most part such checks are done, but I think the level of paranoia was not high enough. That and add the headless support, not sure why tho.... From the readme notes: 3.8.11b1 sqMacOpenGL..c sqMacOpenGLInfo.c sqMacDragDrop.c don't bother with Gestalt, we now drag and drop exists on machines of os-x era. convert to use FSRef Use GetDragAttributes versus bounds in window. Drop WaitNextEvent, dummy up event sqMacUnixFileInterface.c findImageFilterProc use get pathname and last path component to decide if file is image sqMacUnixFileInterface.h sqMacHostWindow.c use QDLocalToGlobalRect in ioPositionOfWindow versus casting rect to point sqMacHostWindow.h sqMacInternetConfiguration.c make FSRef when dealing with ICFileSpec remove some copy pascal to c calls. sqMacJoystickAndTablet.c sqMacMIDI.c sqMacSecurity.c sqMacSerialPost.c sqMacUUID.c sqMacEncoding.c Drop SetImageName() sqMacEncoding.h sqMacFileLogic.h sqMacImageIO.c drop calculateStartLocationForImage() logic sqMacImageIO.h sqMacMain.c drop powerManager logic, not needed, cleanup of statics, cleanup for browser logic cast data right for sprintf(data,"%X",myattr); sqMacMain.h sqMacMemory.c removal of the os-9 swapfile logic watch signed versus unsigned compares sqMacTime.c sqMacTime.h sqMacUIAppleEvents.c use gCurrentVMEncoding, versus system encoding for getting file names drop logic for start of image offset os-9 feature remove PBHSetVolSync() drop LocalToGlobal, use fake mouse location sqMacUIAppleEvents.h sqMacUIClipBoard.c sqmacUIClipBoard.h sqMacUIConstants.h sqMacUIEvents.h sqMacUIEventsUniversal.c remove tap power manager event loop callback In MyWindowEventMouseHandler() put back in 10.2.8 versus 10.3.9 logic sqMacUIMenuBar.h sqMacUUIMenuBarUniversal.c sqMacUnixCommandLineInterface.c sqMacUnixExternalPrims.c sqMacWindow.h sqMacWindowUniversal.c In ioSetFullScreenActual, use QDLocalToGlobalRect, versus casting to points In copy124BitsTheHardWay little versus bit endian for making gworld sqMacNSPluginUILogic.c use QDSwapPort versus get/set port for performance reasons In ioShowDisplayOnWindow avoid double get of windowPort In recordMouseEvent use QDGlobalToLocalPoint sqMacNSPlugin.c Make windowActive = 1, not sure why this didn't fail in the past... Alter GetTempFSSpec to return both fsspec and FSRef 3.8.10b16 Update info.plists and xcode projects on SVN Drop #ifdefs for this below, we always compile on MACH now, versus CodeWarrior Certain end of CodeWarrior OS-9 Support. #if defined (__APPLE__) && defined(__MACH__) #if defined I_AM_CARBON_EVENT #if defined MINIMALVM (sorry Dan) sqMacOpenGL..c sqMacOpenGLInfo.c sqMacWindowPlugin.c sqMacImageIO.c sqMacMain.c sqMacMemory.c sqMacTime.c sqMacUIAppleEvents.c sqMacWindowsUniversal.c sqPlatformSpecific.c sqMacNSPlugin.c osExport.c Export GetUIToLock api so we can callback from FFI on browser plugin UI thread. sqConfig.h Remove #ifdefs above and references to Codewarrior 8 sqmacMain.c change logic to support Sophie Browser plugin SOPHIEVI change logic to get bundle by bundle id when a browser plugin so we can get bundle info.plist for startup logic and when we ask for VM version info remove the gThreadManager OS-9 cooperative thread logic sqMacUIAppleEvents.c browser plugin VM should not Launch Squeak VM on double click. sqMacUIEventsUniversal.c add getUIToLock() logic for Browser plugin support sqMacUnixExternalPrims.c Browser plugin loading bundles support (for squeakland and Sophie) sqMacWindowUniversal.c Fiddle with CGImageCreate 32 host endiness issues Fix cursors data/mask on intel info.plist for Squeak VM Universal sqmacNSPluginUILogic.c complete refactor of bits for Squeak browser plugin, put squeak VM stuff in seperate file sqMacNSPlugin.c complete refactor of bits for Squeak browser plugin, move squeak stuff to other file. Make work with unix file names -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev March 2006