Vm-dev January 2010

vm-dev@lists.squeakfoundation.org

28 participants
47 discussions

Mac VM 5.3b1 ships
by John M McIntosh 11 Jan '10

11 Jan '10

I've push the latest cocoa based 32/64bit macintel/powerpc VM 5.3b1 Various users pointed out that option-arrow key did not work, plus there was a crash due to a draw locking issue. I spent a few hours looking into the complexity of keyboard management in Cocoa and left-overs from NeXTStep. I likely still have more hours of testing for extended keyboards but it should now work for the enter/return, esc, delete, page up/down, and arrows keys with the various combinations of shift, function, control, option, and apple. The crash was related to a race condition in the os-x drawing logic. I refactored it to a more simplistic code base shared between the iPhone and OSX. As a bonus the non-complex code is actually faster than the Carbon Drawing logic found in the 3.x/4.x series of VMs, and much faster than the previous complex/broken logic. Look for the folder experimental/64bits in the regular places found via http://smalltalkconsulting.com/squeak.html or ftp://ftp.smalltalkconsulting.com/experimental/64bit/32bitImage*64bitVM/ or http://homepage.mac.com/johnmci/FileSharing.html click to experimental, 64bits , 32bitImage*64bitVM file name: Squeak 64-32 5.3b1.app.zip This *beta* VM runs in 32 or 64 bit mode on macIntel, and 32bit mode on powerpc to support Squeak 32bit images. Includes the freetype plugin, and requires os-x 10.5 or higher. -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================

1 0

usage of bitShift: in VMMaker & plugins
by Nicolas Cellier 08 Jan '10

08 Jan '10

Folks, browsing generated code I see ugly things like this one: byte = jpegDecodeValueFromsize(dcTable, dcTableSize); if (byte < 0) { return interpreterProxy->primitiveFail(); } if (byte != 0) { bits = getBits(byte); /* begin scaleAndSignExtend:inFieldWidth: */ if (bits < ((((byte - 1) < 0) ? ((usqInt) 1 >> -(byte - 1)) : ((usqInt) 1 << (byte - 1))))) { byte = (bits - (((byte < 0) ? ((usqInt) 1 >> -byte) : ((usqInt) 1 << byte)))) + 1; goto l1; } else { byte = bits; goto l1; } l1: /* end scaleAndSignExtend:inFieldWidth: */; } You would never write such code by hand because it costs a useless runtime test ()?: And you know very well that byte > 0 when engaging the bitShift: operation, so it should be: byte = jpegDecodeValueFromsize(dcTable, dcTableSize); if (byte < 0) { return interpreterProxy->primitiveFail(); } if (byte != 0) { bits = getBits(byte); /* begin scaleAndSignExtend:inFieldWidth: */ if (bits < ((usqInt) 1 << (byte - 1))) { byte = (bits - ((usqInt) 1 << byte)) + 1; goto l1; } else { byte = bits; goto l1; } l1: /* end scaleAndSignExtend:inFieldWidth: */; } Of course, the generator cannot guess that easily when feeded with a bitShift: and a non literal argument... That's why we should use explicit << and >> directly in slang when we know the argument will be positive. I did that for example in http://bugs.squeak.org/view.php?id=7109 but see it could be generalized... ... Of course with due care! Just my 2¢ tip of the day Nicolas

4 5

We want Cog! We want Cog! And some political blabbering...
by Göran Krampe 08 Jan '10

08 Jan '10

Hi all! <rant level=5> Ok, so I was sitting in my car on the way to work this morning, thinking. Well, ok, to be honest I was first sitting up half night surfing about VMs and implementation techniques, etc, for fun, as usual. :) And chatting on a swedish channel about VM tech, LLVM, Parrot, JITs, Forth implementations yadda yadda etc. So... I have two questions: 1. When do we get the first Cog? :) We already know there is a working version. 2. Why is Cog not being developed in an open repo? Perhaps someone already explained #2, but then I forgot. Since Cog is meant to be released as open source (I presume) then I don't get why we can't have access to the work in progress. Or is the decision not really made yet? Is Teleplace going to decide later? Or is Teleplace worried of "tech theft" or something similar? I don't get it. And please don't give us the: "It is too early, we want it to be juuuust right before releasing it to all you people who don't understand these kind of things and just mess it all up if you, eh, get read access to it... ehm." Let me be utterly frank about some "politics" here: You Andreas is putting a whole lot of new energy into Squeak.org, which is a blessing for sure! I love that. But I can't help wondering slightly over the "politics" behind that effort - your time is NOT cheap - I am pretty certain you are important within Teleplace. My guess is that Squeak tech is very important to Teleplace and you really need Squeak.org (and it's cousins) to be live and well etc, and given the rise of Pharo, where you don't have anything near the same influence, I also presume this issue suddenly got "critical". ;) This is all fine and perfectly natural btw, don't read anything suspicious into what I am writing (sorry for the bad english). Now, given all this - why are you holding Cog under a cloak? We all know how open source works and we all know that One Man (although his name is Eliot) can't match the strength of a strong open source team. You need to build a community/team around Cog, you can't just dump it on us and then expect all the Igors and Davids etc to dig in and learn it! It doesn't work that way, it takes time to grow a team and every day that Cog is under wraps is another day wasted growing that team. </rant> Phew. :) regards, Göran PS. Slightly annoyed with the "Sure, we like open source, ehm, when other people do it..." and hoping there is some other perfectly good and viable explanation for the secrecy.

3 3

3.11.3.2135 for Unix
by Ian Piumarta 05 Jan '10

05 Jan '10

Dear Unix Squeakers, 3.11.3.2135 for Unix is now available at squeakvm.org/unix (sources plus binaries for MacOS/Darwin, Linux, NetBSD, FreeBSD and Solaris, on Intel and, occasionally, PowerPC hardware). Some important stuff has changed in this release (see notes below) so please consider it a BETA release even though there is no beta tag on the version number. As usual: comments, suggestions, bug fixes (as whole files please *not* diffs) and any unwanted vinyl LPs sitting in your attic can be sent directly to me. Thanks. Enjoy! Ian Changes since 3.10-6: Numbering scheme changed to X.Y.Z.R where X.Y.Z = VMMaker version used to generate the Interpreter and R is the subversion revision of the platform support code. VMMaker and plugins updated to most recent versions; generated source regenerated: Balloon3D-Plugins-ar.4.mcz DBus-Plugin-bf.32.mcz GStreamer-Plugins-JMM.16.mcz Kedama-Plugins-yo.1.mcz Rome-Plugin-yo.39.mcz OSProcessPlugin-dtl.16.mcz VMConstruction-Plugins-AioPlugin-dtl.9.mcz VMConstruction-Plugins-XDisplayControlPlugin-dtl.7.mcz VMMaker-dtl.138.mcz Build system rewritten using CMake instead of autotools. Big THANK-YOU to Bert, Subbu and Dave Lewis for providing invaluable feedback. Any plugin can now be turned off in configure with the option '--without-<pluginName>'. Experimental PulseAudio driver (vm-sound-pulse) added, thanks to Derek O'Connell. Default is still OSS on Linux; if you want pulse you have to ask for it in one of the usual ways. FFI tweaked on Linux to avoid a potential FPU stack overflow. ClipboardExtendedPlugin will no longer try to build if you don't have X11 development headers and libs. FileCopyPlugin made resilient to EINTR. VM renamed to 'squeakvm'. 'squeak' is now a script that looks for 'squeakvm', figures out if you need any obscure command-line options, and then launches 'squeakvm' for you. This in turn allows... Plugin search stretegy rationalised and simplified. Default location is now the executable directory (where 'squeakvm' is installed). The -plugins argument can be a colon-separated list of locations to search, just like PATH. Plugins are named 'so.plugin' to make interference with FFI libraries less likely. NOTE: FFI libraries are no longer searched for in every single nook and cranny, which means you might have to set LD_LIBRARY_PATH (or your local equivalent) for FFI-based programs to work properly. Failure to load a plugin because of unresolved symbols is now always reported on the terminal. Another script 'squeak.sh' is provided for launching squeak from a menu. This isn't finished and suggestions are welcome. Default path and text encoding is now UTF-8. X11 driver no longer reports SHIFT pressed when CAPS LOCK engaged. SerialPlugin now supports named ports. All occurences of dprintf() renamed to avoid a name conflict with a less-than-stellar decision by POSIX to introduce a standard function of the same name. RomePlugin is no longer built if cairo was built without Pango support. iconv_open() no longer spams you with thousands of error messages if you are missing conversion tables. Valid unicode characters that have no corresponding keycode are reported with keycode zero to avoid an invalid index error in the image. XImmPlugin should now work on MacOS X/Darwin too. Sound driver on MacOS X now works on Intel as well as PowerPC. Various changes to make the VM compile and run on GNU/Linux running on DEC Alpha. vm-sound-Sun now uses asynchronous i/o. Various changes to fix compilation on Solaris.

3 7

B3DAcceleratorPlugin and OpenGL ES
by K. K. Subramaniam 05 Jan '10

05 Jan '10

Hi, Is there any image and plugin supports Balloon Canvas accelerated by OpenGL ES 2.0 library? The B3DAcceleratorPlugin only works with OpenGL and not ES 2.0. I am wondering if anyone has already ported the code to ES 2.0 (say on iPhone, Maemo etc.) TIA .. Subbu

1 0

[ANN] SystemTracer on SqueakSource
by David T. Lewis 03 Jan '10

03 Jan '10

I have created a SqueakSource project called SystemTracing to serve as a repository for the various SystemTracer implementations that have appeared in and around Squeak images since Squeak 1.22. The intent is to make these implementations loadable in modern images, and to provide a Monticello repository for new development. MCHttpRepository location: 'http://www.squeaksource.com/SystemTracing' user: 'squeak' password: 'squeak' All classes and package extension methods have been recategoried to SystemTracing but are otherwise unmodified to the extent possible. The recategorizing was done by hand-editing the original change sets. The archive includes historical versions that appeared in early Squeak images, as well as loadable versions of the SystemTracer2 that is catalogued on SqueakMap. There is an entry for the SystemTracer64 that was used to produce the first 64-bit Squeak image, and recent updates that will enable 64-bit images to be traced from modern Squeak and Pharo images (there is more work to be done here). Developer access is provided for Stephen Pair, John M McIntosh, Andreas Raab, Eliot Miranda, Jannik Laval, Ian Piumarta. If anyone else should be added as developer, please speak up. All future contributions to this archive are required to be MIT licensed. Dave

1 0

Gentoo ebuild fails
by Bert Freudenberg 01 Jan '10

01 Jan '10

FYI - not sure who is maintaining this: http://bugs.gentoo.org/show_bug.cgi?id=298684 - Bert -

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev January 2010