Vm-dev May 2010

vm-dev@lists.squeakfoundation.org

22 participants
59 discussions

[commit] r2207 - CMakeLists builds npsqueak; merge latest npsqueak changes from Bert
by commits＠squeakvm.org 13 May '10

13 May '10

Author: piumarta Date: 2010-05-13 13:36:19 -0700 (Thu, 13 May 2010) New Revision: 2207 Modified: trunk/platforms/unix/CMakeLists.txt trunk/platforms/unix/ChangeLog trunk/platforms/unix/cmake/squeak.in trunk/platforms/unix/npsqueak/README.npsqueak trunk/platforms/unix/npsqueak/npsqueak.c trunk/platforms/unix/npsqueak/npsqueakregister.in trunk/platforms/unix/npsqueak/npsqueakrun.in Log: CMakeLists builds npsqueak; merge latest npsqueak changes from Bert Modified: trunk/platforms/unix/CMakeLists.txt =================================================================== --- trunk/platforms/unix/CMakeLists.txt 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/CMakeLists.txt 2010-05-13 20:36:19 UTC (rev 2207) @@ -1,6 +1,6 @@ # Yes, it's true: you can write makefiles in COBOL. # -# Last edited: 2010-04-09 04:23:35 by piumarta on ubuntu +# Last edited: 2010-05-13 12:54:20 by piumarta on emilia-2.local PROJECT (squeak) @@ -137,6 +137,8 @@ BUILD_PLUGINS () +ADD_SUBDIRECTORY (${unix}/npsqueak) + SET (cmake ${unix}/../../cmake) IF (EXISTS ${cmake}/Packages.cmake) Modified: trunk/platforms/unix/ChangeLog =================================================================== --- trunk/platforms/unix/ChangeLog 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/ChangeLog 2010-05-13 20:36:19 UTC (rev 2207) @@ -1,3 +1,20 @@ +2010-05-13 Ian Piumarta <com -dot- gmail -at- piumarta (backwards)> + + * CMakeLists.txt: Add subdir npsqueak. + + * npsqueak.c: Cosmetic improvements to error messages. + + * npsqueak/README.npsqueak: Updated for CMake. + + * npsqueak/npsqueakrun.in: New version from Bert. + + * npsqueak/npsqueakregister.in (NPSQUEAK_SO): Construct from + substituted cmake variables. + + * cmake/squeak.in (PATH): realpath() prepends dir to link target + iff target is relative. Only echo environment if SQUEAK_DEBUG is + nonzero. + 2010-05-11 Ian Piumarta <com -dot- gmail -at- piumarta (backwards)> * vm-display-X11/sqUnixX11.c: Add option -closequit and Modified: trunk/platforms/unix/cmake/squeak.in =================================================================== --- trunk/platforms/unix/cmake/squeak.in 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/cmake/squeak.in 2010-05-13 20:36:19 UTC (rev 2207) @@ -3,7 +3,7 @@ # Launch squeakvm from the command line or a menu script, with a good # plugin path, text encodings and pulseaudio kludge # -# Last edited: 2010-04-29 02:52:25 by piumarta on margaux1 +# Last edited: 2010-05-13 13:27:12 by piumarta on emilia-2.local PATH=/usr/bin:/bin @@ -14,7 +14,9 @@ dir=`cd "${dir}" && pwd -P` path=`basename "${path}"` path=`ls -l "${dir}/${path}" | sed 's,.* -> ,,'` - path="${dir}/${path}" + if test `expr "${path}" : "/"` -eq 0; then + path="${dir}/${path}" + fi done if test -d "${path}"; then (cd "${path}" && pwd -P) @@ -91,6 +93,13 @@ export LC_ALL fi +# debug output + +if test "0$SQUEAK_DEBUG" -gt "0"; then + set | fgrep SQUEAK_ + set -x +fi + # run the vm set | fgrep SQUEAK_ Modified: trunk/platforms/unix/npsqueak/README.npsqueak =================================================================== --- trunk/platforms/unix/npsqueak/README.npsqueak 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/npsqueak/README.npsqueak 2010-05-13 20:36:19 UTC (rev 2207) @@ -4,13 +4,12 @@ Author: Bert Freudenberg -Version: @VM_VERSION@ -Last edited: 2005-02-28 15:50:31 by jens on impara.de +Last edited: 2010-05-11 17:30:59 by bert ================================================================ Files: - Makefile + CMakeLists.txt README.npsqueak npsqueak.c npunix.c (modified copy from Netscape Plugin SDK) @@ -23,15 +22,15 @@ The Squeak plugin should be automatically compiled and installed with the normal squeak build process. This installs squeak into (probably) - imgdir=/usr/local/lib/squeak + squeakdir=/usr/local/lib/squeak - You will need an image. Copy it to + You will need an image. Copy it (or symlink) to - ${imgdir}/SqueakPlugin.image + ${squeakdir}/SqueakPlugin.image A good place to look for an image suited as plugin is - http://squeakland.org/plugin/installers/files.html + http://squeakland.org/download To register the plugin run ${imgdir}/npsqueakregister. It looks for various browsers @@ -39,17 +38,16 @@ your browser is not found, hack the npsqueakregister script or symlink the plugin manually: - ln -s ${imgdir}/@VM_VERSION@/npsqueak.so \ + ln -s ${squeakdir}/@VM_VERSION@/npsqueak.so \ /usr/lib/mybrowser/plugins/ - Exit and restart the browser. Check if the Squeak plugin was recognized by selecting the Help>>About Plugins menu item. Open the html file in the test/ directory. This should activate the Squeak plugin. Since this is the first time the plugin is executed, a .npsqueak folder will be created in your - home directory and the plugin image is installed into it. + home directory. ================================================================= @@ -63,41 +61,50 @@ DOCUMENTATION -* On the first run for each user, the npsqueakrun will create a directory - structure like this: +* How the plugin works: + - npsqueak.so needs to be symlinked where the browser expects plugins + - when the browser wants to display a page with an embedded Etoys + project, it launches the plugin + - the plugin executes the "npsqueakrun" script, passing some parameters + to establish a communication channel between the plugin and squeak + - that script runs the Squeak VM with an image, instructing the VM to + redirect the output into the plugin's window + - the image then loads the project + +* On the first run for each user, npsqueakrun will create a directory + structure like this (unless overridden by variables, see below): + $(HOME)/ | +--- .npsqueak/ - | - +--- SqueakPlugin.image - | - |--- *.image - | - +--- secure/ - | - +--- My Squeak/ + | | + | +--- secure/ + | + +--- Squeak/ - $(imgdir)/ - | - +--- SqueakPlugin.image - * Environment Variables: SQUEAK_USERDIR Identifies the sandbox directory, e.g., a place which can be accessed from any (potentially unsafe) Squeaklet. - * path must be absolute + * Default: "$(HOME)/Squeak" - * Default: "~/.npsqueak/My Squeak" - + SQUEAK_SECUREDIR + Identifies the directory to store private information + like signing keys. + * should not be accessible by other users + + * Default: "$(HOME)/.npsqueak/secure" + NPSQUEAK_DEBUG - logs debug info to /tmp/npsqueak.log - Default: don't log + If set, logs debug info to /tmp/npsqueak.log + * Default: unset + * Mimetypes: application/x-squeak-source:sts:Squeak source @@ -121,12 +128,25 @@ * loop on read() and write() until all data is sent and received. * make plugin not wait for browser when the url is loaded into a target. -* use autoconf ================================================================= Change log +May 2010: + * use CMake + * default to ~/Squeak + * do not copy image to ~/.npsqueak anymore + +Aug 2009: + * (ikp) BSD-friendly Makefile + +Sep 2007: + * (bf) use "find -H" in npsqueakregister + +Sep 2006: + * (ikp) use libtool to build + Jan 2005: * npsqueak checks for image file in system and home dir * modified npsqueakrun, runs with bash, other shells untested Modified: trunk/platforms/unix/npsqueak/npsqueak.c =================================================================== --- trunk/platforms/unix/npsqueak/npsqueak.c 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/npsqueak/npsqueak.c 2010-05-13 20:36:19 UTC (rev 2207) @@ -4,7 +4,7 @@ * * Author: Bert Freudenberg * - * Last edited: 2005-03-17 12:15:48 by piumarta on squeak.hpl.hp.com + * Last edited: 2010-05-11 20:31:45 by bert * * History: * Jan 2005 - looking for image and npsqueakrun in system and home dir @@ -283,7 +283,7 @@ char* bin_dir_v[PATH_MAX]= {user_bin_dir, SYSTEM_BIN_DIR"/"}; if (findFileInPaths(plugin->vmName, NPSQUEAKRUN, 2 , bin_dir_v) == 0){ - fprintf(stderr, "Squeak Plugin: npsqueakrun not found!\n"); + fprintf(stderr, "Squeak Plugin: "NPSQUEAKRUN" not found!\n"); return NPERR_GENERIC_ERROR; } } @@ -298,7 +298,7 @@ failureUrl); plugin->failureUrl= NPN_StrDup(failureUrl); }else { - fprintf(stderr, "Squeak Plugin: no failure URL: \n"); + fprintf(stderr, "Squeak Plugin: no failure URL.\n"); return NPERR_GENERIC_ERROR; } } Modified: trunk/platforms/unix/npsqueak/npsqueakregister.in =================================================================== --- trunk/platforms/unix/npsqueak/npsqueakregister.in 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/npsqueak/npsqueakregister.in 2010-05-13 20:36:19 UTC (rev 2207) @@ -7,7 +7,7 @@ # Rerun after you installed a new browser! # Parameters: -u unregister -NPSQUEAK_SO=@NPSQUEAK_SO@ +NPSQUEAK_SO=@prefix@/@plgdir@/npsqueak.so BROWSERS="netscape mozilla mozilla-firefox opera MozillaFirefox/lib" BROWSER_DIRS="/usr/local/lib /usr/lib /opt" SYSTEM_BROWSER_DIR="/usr/lib/browser-plugins" @@ -19,7 +19,7 @@ echo "Registers the Squeak browser plugin for known browsers" echo " -u unregister plugin" echo " PLUGIN absolute path to npsqueak.so" - echo " (default is @NPSQUEAK_SO@)" + echo " (default is $NPSQUEAK_SO)" exit } Modified: trunk/platforms/unix/npsqueak/npsqueakrun.in =================================================================== --- trunk/platforms/unix/npsqueak/npsqueakrun.in 2010-05-12 00:02:20 UTC (rev 2206) +++ trunk/platforms/unix/npsqueak/npsqueakrun.in 2010-05-13 20:36:19 UTC (rev 2207) @@ -2,19 +2,39 @@ # File: npsqueakrun # Author: Bert Freudenberg -# Last edited: 2005-02-28 18:42:20 by jens on impara.de # Description: Script to start the Squeak VM binary # from the Browser Plugin (npsqueak.so) -# If necessary, sets up $HOME/.npsqueak/* first. +# If necessary, sets up $HOME/.npsqueak first. +# Arguments: +# $0 = @prefix@/@imgdir@/npsqueakrun +# $1,$2 = -display :0.0 +# $3-$5 = -browserPipes 9 12 +# $6 = /path/to/some.image +# $7 = '' +# $8,$9 = key1 value1 (tags as given in the html page) +# etc. = key2 value2 -VM_VERSION=@VM_VERSION@ -imgdir=@imgdir@ - -IMAGE=${imgdir}/SqueakPlugin.image -VM=squeak -VMOPTIONS=" -mmap 1024 -vm-display-x11 -swapbtn " +VM=@prefix@/@bindir@/squeak +VMOPTIONS= LOGFILE=/tmp/npsqueak.log +realpath () { + path="$1" + while test -L "${path}"; do + dir=`dirname "${path}"` + dir=`cd "${dir}" && pwd -P` + path=`basename "${path}"` + path=`ls -l "${dir}/${path}" | sed 's,.* -> ,,'` + done + if test -d "${path}"; then + (cd "${path}" && pwd -P) + else + dir=`dirname "${path}"` + base=`basename "${path}"` + (cd "${dir}" && echo "`pwd -P`/${base}") + fi +} + pdebug() { if [ ! -z $NPSQUEAK_DEBUG ]; then @@ -31,18 +51,6 @@ fi } -# is there a clever way to determine if you are local? -if [ $DISPLAY = ":0.0" ]; then - VMOPTIONS=$VMOPTIONS" -xshm " - pdebug "using x shared memory" -else - pdebug "disabling x shared memory, because we are not local" -fi - -if [ -e /usr/bin/squeakvm ]; then - VM=squeakvm -fi - ensuredir() { if [ ! -d "${1}" ] ; then @@ -57,49 +65,8 @@ fi } -ensurefile() +parameter_to_array() { - if [ ! -e "${1}" ] ; then - if cp "${2}" "${1}" ; then - pdebug "Created ${1}" - else - perror "Could not create ${1}" - if [ ! -r "${2}" ] ; then - perror "because ${2} is missing" - fi - fi - fi -} - -ensureimage() -{ - if [ `dirname "${1}"` = "${HOME}/.npsqueak" ] ; then - IMAGE=$1 - else - IMAGE="${HOME}/.npsqueak/"`basename "$1"` - ensurefile "$IMAGE" "$1" - fi -} - -ensure() -{ - if [ `dirname "${1}"` = "${HOME}/.npsqueak" ] ; then - IMAGE=$1 - else - IMAGE="${HOME}/.npsqueak/"`basename "$1"` - ensurefile "$IMAGE" "$1" - fi -} - -if [ -e "${HOME}/.npsqueak" ] ; then - pdebug "don't create MySqueak link on desktop," - pdebug "because ${HOME}/.npsqueak exists" -else - pdebug "create MySqueak link on desktop" - ln -s "${HOME}/.npsqueak/My Squeak" "${HOME}/Desktop/My Squeak" -fi - -parameter_to_array(){ i=0 for p in "$@"; do i=$(( $i + 1 )) @@ -107,27 +74,17 @@ done } -ensuredir "${HOME}/.npsqueak" -ensuredir "${HOME}/.npsqueak/secure" 700 -if [ ! -z "$SQUEAK_USERDIR" ]; then - pdebug "using SQUEAK_USERDIR" - pdebug "My Squeak -> $SQUEAK_USERDIR" -else - pdebug "My Squeak -> ${HOME}/.npsqueak/My Squeak" - ensuredir "${HOME}/.npsqueak/My Squeak" -fi +[ -z "$SQUEAK_USERDIR" ] && export SQUEAK_USERDIR="$HOME/Squeak" +[ -z "$SQUEAK_SECUREDIR" ] && export SQUEAK_SECUREDIR="$HOME/.npsqueak/secure" -if [ -z "$1" ]; then - pdebug "start squeak direct" - ensureimage $IMAGE - $VM $VMOPTIONS $IMAGE -else - pdebug "start squeak from plugin" - IMAGE="${6}" - ensureimage $IMAGE - parameter_to_array "$@" - parameter[6]=$IMAGE - $VM $VMOPTIONS "${parameter[@]}" -fi +ensuredir "$SQUEAK_USERDIR" +ensuredir "$SQUEAK_SECUREDIR" 700 +# resolve image symlink so Smalltalk imagePath is correct +IMAGE="${6}" +parameter_to_array "$@" +parameter[6]=`realpath "$IMAGE"` + +$VM $VMOPTIONS "${parameter[@]}" +

1 0

Re: Freetype plugin archive
by David T. Lewis 12 May '10

12 May '10

On Sun, May 02, 2010 at 08:55:33AM -0700, Michael Rueger wrote: > On 5/1/10 11:31 AM, Bert Freudenberg wrote: > >On 01.05.2010, at 18:31, David T. Lewis wrote: > > >>Do you know if the rest of the stuff in http://source.impara.de/freetype/ > >>still has value (or has it been superceded by Freetype Plus)? > > I would think so, that's why there wasn't any further development on the > freetype plugin. > The current repositories are of somewhat historical value for e.g. the > Sophie builds. I have opened a new SqueakSource project for the Freetype plugin (FT2Plugin). MCHttpRepository location: 'http://www.squeaksource.com/FreetypePlugin' user: '' password: '' This contains copies of all the Freetype-Plugin entries from the archival repository at http://source.impara.de/freetype, as well as one additional update from Mantis 7521. Project administrators are Andy Tween, Bert Freudenberg and me. Developers are everyone whose author initials appear in the original archive. VMMaker class>>updateFromServer should now load FT2Plugin from the new FreetypePlugin repository. A similar update will be needed for ConfigurationOfVMMaker for Pharo. For now the original package name Freetype-Plugin has been retained, although I note that this conflicts with the current packaging of Freetype in SqS/Freetype Plus, so a change to the package name may be in order. FT2Plugin contains dependencies on Freetype. This is the reason that I moved it to a separate SqueakSource project rather than include it in the SqS/VMMaker repository. Dave

4 12

[ANN] DTrace for Squeak/Pharo
by Adrian Lienhard 12 May '10

12 May '10

Now there's also a DTrace provider for the Squeak VM: http://www.adrian-lienhard.ch/blog?dialog=smalltak-meets-dtrace DTrace is a powerful, kernel-based dynamic tracing framework. With the new provider one can analyze the execution of Pharo/Squeak applications. My post is an introduction that should get you quickly started. I've attached a pre-compiled VM for OS X. Hope you like it :) Cheers, Adrian ___________________ http://www.adrian-lienhard.ch/

3 4

Where is the heap?
by Javier Pimás 12 May '10

12 May '10

Hi, I'm working in memory management of SqueakNOS platform. Currently we use a static size heap, and we'll probably have to change that. I read in the blue book that the heap is implemented as a *contiguous* chunk of memory, is the heap done that way in squeak? I looked how it's implemented in unix and saw that sqAllocateMemory uses mmap. I don't have experience with this function, but I wonder if it means that the heap is allocated in a virtual contiguous chunk that gets mapped to some place in disk. Or maybe it just reserves contiguous virtual addresses that are in different blocks of real memory? The alloc line is this, (sqUnixMemory.c) if (MAP_FAILED == (heap= mmap(0, heapLimit, MAP_PROT, MAP_FLAGS, devZero, 0))) Also, other question is what is this mapping to /dev/zero? Then reading from the heap wouldn't return allways zero? Is it because this mmap isn't actually backed by any file? Does mmap reserve heapLimit bytes in real memory? I mean, if you write to the mapped address, you write to real memory, device memory, or both? Lastly, growing the heap gets to this: if (MAP_FAILED == mmap(base, newDelta, MAP_PROT, MAP_FLAGS | MAP_FIXED, devZero, heapSize)) which doesn't alter heap start, so it has to be that when resizing the heap the only thing that happens is that some memory is reserved and mapped to the ending address of it. Well as you can see I'm rather confused so any help is wellcome, thanks for reading. Javier. -- Javier Pimás Ciudad de Buenos Aires

5 6

[commit] r2206 - quit when window closed if option -closequit or SQUEAK_CLOSEQUIT set
by commits＠squeakvm.org 12 May '10

12 May '10

Author: piumarta Date: 2010-05-11 17:02:20 -0700 (Tue, 11 May 2010) New Revision: 2206 Modified: trunk/platforms/unix/ChangeLog trunk/platforms/unix/vm-display-X11/sqUnixX11.c Log: quit when window closed if option -closequit or SQUEAK_CLOSEQUIT set Modified: trunk/platforms/unix/ChangeLog =================================================================== --- trunk/platforms/unix/ChangeLog 2010-04-29 17:17:54 UTC (rev 2205) +++ trunk/platforms/unix/ChangeLog 2010-05-12 00:02:20 UTC (rev 2206) @@ -1,3 +1,8 @@ +2010-05-11 Ian Piumarta <com -dot- gmail -at- piumarta (backwards)> + + * vm-display-X11/sqUnixX11.c: Add option -closequit and + environment variable SQUEAK_CLOSEQUIT. + 2010-04-13 Ian Piumarta <com -dot- gmail -at- piumarta (backwards)> * doc/RELEASE_NOTES_4.0.3.2202: Release 4.0.3.2202. Modified: trunk/platforms/unix/vm-display-X11/sqUnixX11.c =================================================================== --- trunk/platforms/unix/vm-display-X11/sqUnixX11.c 2010-04-29 17:17:54 UTC (rev 2205) +++ trunk/platforms/unix/vm-display-X11/sqUnixX11.c 2010-05-12 00:02:20 UTC (rev 2206) @@ -27,7 +27,7 @@ /* Author: Ian Piumarta <ian.piumarta(a)squeakland.org> * - * Last edited: 2010-04-09 15:04:50 by piumarta on ubuntu + * Last edited: 2010-05-11 16:58:09 by piumarta on emilia-2.local * * Support for more intelligent CLIPBOARD selection handling contributed by: * Ned Konz <ned(a)bike-nomad.com> @@ -327,6 +327,7 @@ int noTitle= 0; int fullScreen= 0; int iconified= 0; +int closeQuit= 0; int withSpy= 0; @@ -3383,8 +3384,8 @@ XFree((void *)wmHints); } - /* tell the WM that we do not want to be destroyed from the title bar close button */ - { + if (!closeQuit) { + /* tell the WM that we do not want to be destroyed from the title bar close button */ wmProtocolsAtom= XInternAtom(stDisplay, "WM_PROTOCOLS", False); wmDeleteWindowAtom= XInternAtom(stDisplay, "WM_DELETE_WINDOW", False); XSetWMProtocols(stDisplay, stParent, &wmDeleteWindowAtom, 1); @@ -5828,6 +5829,7 @@ printf("\nX11 <option>s:\n"); printf(" -browserWindow <wid> run in window <wid>\n"); printf(" -browserPipes <r> <w> run as Browser plugin using descriptors <r> <w>\n"); + printf(" -closequit quit immediately when the main window is closed\n"); printf(" -cmdmod <n> map Mod<n> to the Command key\n"); printf(" -compositioninput enable overlay window for composed characters\n"); printf(" -display <dpy> display on <dpy> (default: $DISPLAY)\n"); @@ -5890,6 +5892,7 @@ if (getenv("SQUEAK_NOXDND")) useXdnd= 0; if (getenv("SQUEAK_FULLSCREEN")) fullScreen= 1; if (getenv("SQUEAK_ICONIC")) iconified= 1; + if (getenv("SQUEAK_CLOSEQUIT")) closeQuit= 1; if (getenv("SQUEAK_MAPDELBS")) mapDelBs= 1; if (getenv("SQUEAK_SWAPBTN")) swapBtn= 1; if ((ev= getenv("SQUEAK_OPTMOD"))) optMapIndex= Mod1MapIndex + atoi(ev) - 1; @@ -5921,6 +5924,7 @@ else if (!strcmp(arg, "-swapbtn")) swapBtn= 1; else if (!strcmp(arg, "-fullscreen")) fullScreen= 1; else if (!strcmp(arg, "-iconic")) iconified= 1; + else if (!strcmp(arg, "-closequit")) closeQuit= 1; #if !defined (INIT_INPUT_WHEN_KEY_PRESSED) else if (!strcmp(arg, "-nointl")) initInput= initInputNone; #else

4 3

Set and get values from the VM AND the image
by Mariano Martinez Peck 10 May '10

10 May '10

Hi folks. I need to intercept all messages. For such purpose, I modified Interpreter >> normalSend like this: normalSend "Send a message, starting lookup with the receiver's class." "Assume: messageSelector and argumentCount have been set, and that the receiver and arguments have been pushed onto the stack," "Note: This method is inlined into the interpreter dispatch loop." | rcvr | self inline: true. self sharedCodeNamed: 'normalSend' inCase: 131. rcvr := self internalStackValue: argumentCount. (self isIntegerObject: rcvr) ifFalse: [ self doSomething: rcvr. ]. lkupClass := self fetchClassOf: rcvr. receiverClass := lkupClass. self commonSend. But now, from the IMAGE side I would like to enable or disable such intercept. It would be great if from the Image side I can set a value to a boolean and read such value and ask, from the VM side. Do you know if I can do this? Should I use specialObjectArray ? If true, can you give me a hint of how to do it ? Thank you very much. Mariano

2 3

Re: [squeak-dev] Vm crash from FileList
by David T. Lewis 09 May '10

09 May '10

(CC to vm-dev list) Hannes, Thanks for providing the JPEG file. This is Mantis 7523: Opening some JPEG files may crash Windows VM Apparently certain (possibly corrupt) JPEG files can crash the Windows VM, suggesting a problem with error handling in the support code or external libraries. A file that reliably reproduces the problem is attached to the Mantis report. Dave On Sun, May 09, 2010 at 07:19:30AM +0000, Hannes Hirzel wrote: > Dave, > > I have uploaded the jpg file to http://bugs.squeak.org/view.php?id=7523. > > It leads to an immediate crash on MSWindows whereas it displays fine on Ubuntu. > > --Hannes > > On 5/3/10, David T. Lewis <lewis(a)mail.msen.com> wrote: > > On Mon, May 03, 2010 at 03:50:55PM +0000, Hannes Hirzel wrote: > >> Hello > >> > >> I have found a 110kB sample jpg file which causes the MSWindows VM to > >> crash immediately *** whereas it displays the file properly under > >> Ubuntu. > >> > >> I will happily sent the file to anybody interested having a look at > >> this.... > > > > Hannes, > > > > I opened a Mantis report for this. Can you please upload your sample > > jpg file here: http://bugs.squeak.org/view.php?id=7523 > > > > Thanks, > > Dave > >

1 0

Build squeak vm screencast updated (with FT2Plugin and Alien)
by laurent laffont 09 May '10

09 May '10

on pharocasts: http://pharocasts.blogspot.com/2010/03/build-squeak-vm-on-linux.html Feedback always appreciated :) Cheers, Laurent Laffont

1 0

About CompactClasses and their future [WAS] how many free bits we have in Object Header?
by Mariano Martinez Peck 06 May '10

06 May '10

Hi folks. I was just trying to understand this optimization of the Compact Classes. If I understood correctly, most objects needs a second word there they have 30 bits of OOP of their class and 2 for the header type. There are classes that have a LOT of instances. For those cases we have compact classes where their instances have a number using those 5 bits in the first word of the header, and thus, they don't need a second header. Then, less memory is used. So...I understood ok ? Now, if that is the situation. I have evaluated Smalltalk compactClassesArray and I see that in the array there are a lot of nils. 16 classes in Squeak and 15 in Pharo. The rest of the 31, are just nil. Now I wonder if it is worth to do anything with that : a) Use 4 bits instead of 5 and free one more bit in the OH b) remove completely CompactClasses as Adrian said c) Let 5 bits but then add more classes (up to 31) so that we can get benefits of that. Do you know if there are side effects of becomingCompact a class ? If I evaluate SpaceTally new printSpaceAnalysis I get something like this: Class code space # instances inst space percent ByteString 2217 71317 3621735 19.3 CompiledMethod 21064 56057 3517772 18.7 Bitmap 3893 2256 3424604 18.2 Array 2478 71108 2198420 11.7 ByteSymbol 959 37896 962465 5.1 ByteArray 2778 317 778297 4.1 MethodDictionary 1819 5964 449252 2.4 MCVersionInfo 1313 10369 414760 2.2 DateAndTime 6974 13793 331032 1.8 WeakArray 827 286 326900 1.7 Association 749 26784 321408 1.7 UUID 1528 10370 248880 1.3 Float 11442 16450 197400 1.1 ClassOrganizer 1398 5964 190848 1.0 ODatedEntry 263 7049 169176 0.9 Time 5530 10369 165904 0.9 Duration 3025 10290 164640 0.9 Date 3167 10289 164624 0.9 LargePositiveInteger 2515 14340 118169 0.6 Metaclass 4250 2876 115040 0.6 Point 6404 8476 101712 0.5 GlyphForm 307 2030 73080 0.4 FreeTypeCacheEntry 536 2238 71616 0.4 SpaceTallyItem 651 2509 60216 0.3 OrderedCollection 5351 2831 56620 0.3 FreeTypeFileInfo 797 321 30816 0.2 FreeTypeFontFamilyMember 1274 537 25776 0.1 Color 21202 1255 25100 0.1 TraitComposition 2651 2013 24156 0.1 Set 3510 1358 21728 0.1 Dictionary 5742 1344 21504 0.1 RemoteString 1533 1236 19776 0.1 SparseLargeTable 2376 2 18708 0.1 WordArray 1821 30 18608 0.1 SoundBuffer 3299 3 18508 0.1 Pragma 1558 862 17240 0.1 Preference 1534 379 16676 0.1 EventHandler 3216 125 14500 0.1 MorphExtension 3249 277 14404 0.1 Character 6428 939 11268 0.1 SortedCollection 1496 462 11088 0.1 AdditionalMethodState 2576 859 10304 0.1 ISOLanguageDefinition 18114 402 9648 0.1 TimeStamp 861 392 9408 0.1 So...for this quick evaluation, maybe we can become compact also the following classes for example: Smalltalk compactClassesArray #( ByteSymbol ByteArray MethodDictionary "Association" "WeakArray" "Metaclass" DateAndTime UUID Time Duration Date "MethodDictionary" ClassOrganizer Metaclass Pragma OrderedCollection RemoteString WordArray Color Character ) do: [:each | (Smalltalk at: each) becomeCompact] Do this make sense ? How could I measure this? I took a Pharo image and compare the memory used by the VM running all test before and after those new compact classes. From the "Activity monitor" I noticed between 3 and 5 MB less used with these new compact classes. Thanks in advance Mariano On Tue, May 4, 2010 at 2:12 PM, Adrian Lienhard <adi(a)netstyle.ch> wrote: > > Hi Mariano, > > I see two solutions > - make the identityHash (even) smaller > - remove compact classes > > The former would probably not be hard to implement (just change some bit > masks and then rehash all sets). I've brought up the idea about the latter > some time ago on this mailing list. The idea is to remove compact classes to > get a larger identityHash (trading memory against speed). After the removal, > only two header types are left and hence you gain a spare bit. I think this > shouldn't be too much work either, but I haven't come around to implementing > it (yet). > > Good luck ;) > Adrian > > On May 4, 2010, at 13:13 , Mariano Martinez Peck wrote: > > >>> > >>> Now...the question is, is there another free bit in the object header ? > I > >>> need another one :( > >> > >> Based on the comment at the bottom of Interpreter>>internalIsImmutable: > >> I suspect that somebody else has their eye on that last available bit > >> also ;-) > >> > >> > > Yes, I know :( But for the moment I don't care to "integrate" my stuff. > I > > just want to play and experiment. And even for that, I need 2 bits :( > > Of course, if I once make something good to integrate, I will have double > > problem :( > > > > So...no solution ? > >

3 3

how many free bits we have in Object Header?
by Mariano Martinez Peck 06 May '10

06 May '10

Hi folks. I think I am using the last free bit in the object header. For the moment, I am using this: ObjectMemory >> initializeObjectHeaderConstants ..... "masks for root and mark bits" MarkBit := 1 bitShift: BytesPerWord*8 - 1. "Top bit" RootBit := 1 bitShift: BytesPerWord*8 - 2. "Next-to-Top bit" AllButMarkBit := WordMask - MarkBit. AllButRootBit := WordMask - RootBit. AllButMarkBitAndTypeMask := AllButTypeMask - MarkBit. UsedObjectBit := 1 bitShift: BytesPerWord*8 - 3. "Next-to-Top bit" AllButUsedBit := WordMask - UsedObjectBit. Now...the question is, is there another free bit in the object header ? I need another one :( Thanks Mariano

3 5

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev May 2010