Vm-dev March 2006

vm-dev@lists.squeakfoundation.org

14 participants
19 discussions

FloatMathPlugin for Croquet
by Andreas Raab 08 Apr '06

08 Apr '06

Hi Guys - I've just committed the changes for the FloatMathPlugin which Croquet uses to get bit-identical cross-platform floating point behavior (with a big thanks to Sun for developing fdlibm ;-) Getting FloatMathPlugin to compile should be simple; getting it to work might be tricky. Because of this there are a bunch of tests with extensive test data to ensure we truly have cross-platform behavior. For efficiency, it's important to note that the "common" floating point operations are not handled by fdlibm (it expects IEEE 754 conforming behavior for the basic operations like add, subtract, multiply, divide) but covers pretty much everything else. What I have found playing with both PPC and Intel FPUs is the following: Even if your FPU is IEEE754 compliant this means very little, as IEEE754 leaves lots of room for interpretation. Most importantly, make sure you check the following two issues: * Internal accuracy. Some FPUs (x87) have an internal accuracy that is larger than 64bits (80bits on Intel). For compliance the FPU *MUST* be set to treat all operations as 64bit (on x87 this means setting the precision to 53 bits, e.g., _PC_53 in the control word). * Fused multiply-add. Some FPUs (PPC) allow for so-called "fused" multiply-add operations which compute the result of "a*b+c" with higher internal accuracy. You must disable fused multiply-add *AT ALL COSTS* (e.g., by using a global compiler switch if available) as the use of fused multiply-add is both, subject to choice by the compiler (meaning that even two different compilers might decide differently where to use it) and is not generally supported on all FPUs. The tests that I've provided come in two flavours: One flavour is just a general compliance test which computes a bunch of numbers hashes them (using MD5) and compares it with a precomputed value. This test is a good quick way to check for compliance. The second flavour uses pre-computed test data so you can actually look at what is computed differently and where. The test data also comes in two flavours, a small data set (10k entries) and a large data set (1000k entries). The small one is a good one early on to find the problematic cases quickly, the large one is good to check that with 1:1000000 chance the code actually does compute the correct values. You need to drop the data sets into the image directory for those tests to be run. Since (so far) I have only tested this on x87 and PPC, I don't know if other FPUs have other issues. If you encounter "wrong" results despite checking for the above two issues you might want to read the manual for your FPU to find out if there are any settings that may affect IEE754 compliance (like fused-madd, or internal accuracy). Cheers, - Andreas PS. And apologies for the temporary SVN mess - I wasn't aware that my SVN client wouldn't interpret the directory I had clicked on as being relative to the URL (and ended up with committing everything to the Cross/plugins directory instead of Cross/plugins/FloatMathPlugin).

3 3

InterpreterProxy>>ioFilename:fromString:ofLength:resolveAliases:?
by Andreas Raab 04 Apr '06

04 Apr '06

This function is news to me - can somebody explain what it's good for, where it should be used and (most importantly) why we'd export this for plugins but don't use in the VM itself (e.g., no senders)? Thanks, - Andreas

6 20

sq.h changes, grabbed reserved field in sqKeyboardEvent
by John M McIntosh 02 Apr '06

02 Apr '06

In our work with Sophie we became aware of a shortcoming in how keyboard events are created and what data they contain and what data is needed to transform keyboard command on various obscure European keyboards into meaningful interactions within Tweak and other important applications like Sophie. In the past on the macintosh and we believe on Windows the data coming up as a result of a keystroke was key down charCode: a magic number mapping to a key on the keyboard, different between mac/windows pressCode: 0 key char charCode the macroman value of the key pressed pressCode the utf-32 value of the key pressed key up charCode: a magic number mapping to a key on the keyboard, different between mac/windows pressCode: 0 Now the results will be. key down charCode: a magic number mapping to a key on the keyboard, different between mac/windows pressCode the utf-32 value of the key pressed additionalData the macroman value of the key pressed key char charCode the macroman value of the key pressed pressCode the utf-32 value of the key pressed additionalData a magic number mapping to a key on the keyboard, different between mac/windows key up charCode: a magic number mapping to a key on the keyboard, different between mac/windows pressCode the utf-32 value of the key pressed additionalData the macroman value of the key pressed This preserves the existing behaviour, but adds the ability to better support non-usa keyboards and character sets. I will make this change btw weekend baring a more elegant solution. /* keyboard input event */ typedef struct sqKeyboardEvent { int type; /* EventTypeKeyboard */ unsigned int timeStamp; /* time stamp */ int charCode; /* character code in Mac Roman encoding */ int pressCode; /* press code; any of EventKeyXXX */ int modifiers; /* combination of xxxKeyBit */ int utf32Code; /* UTF-32 unicode value */ int additionCharacterData; /* Additional data about charcter for platform specific tweak dispatching */ int windowIndex; /* host window structure */ } sqKeyboardEvent; -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

5 16

Building Unix Squeak VM On Intel Mac
by Todd Blanchard 31 Mar '06

31 Mar '06

I've got a new MacBook and want to build a native version of the Unix VM. I'm stymied by the existence of powerpc assembly in sqUnixQuartz.m Any hints on how to work around this? Thanks, Todd Blanchard

3 5

Can plugins add libraries to unix vm?
by Alan Sibley 28 Mar '06

28 Mar '06

Hey folks, We are working on a plugin using the unix VM for squeak under OSX Tiger. I'd like to have the resulting squeak VM link in some foundation stuff from OSX (CoreFoundation CoreServices...). Is there a way to trigger this using the Makefile.in stuff (tried altering XLDFLAGS, PLIBS, LIBS with no success). It would be nice to have this linkage only occur if adding in our plugin. Thanks in advance, Alan

7 46

All together now ;-)
by Andreas Raab 28 Mar '06

28 Mar '06

Hi Guys - You probably noticed that I was busy working on the latest Croquet VMs - the whole set of issues and changes that I am dealing with right now are for Croquet 1.0. It seems like things are settling down now (at least I have no obvious things on my TODO list any longer) and a consolidated version of VMMaker which includes all the issues discussed here is available at the repository at: http://hedgehog.software.umn.edu:8888/Homebase.html Tim - This (or perhaps the version two weeks from now ;-) is probably a good version for you to merge back into the main line when you find the time for it. There aren't many changes, just what I was dealing with in general (FloatMathPlugin, Tests, CroquetPlugin, SocketPlugin) so integration should be straightforward. Cheers, - Andreas

4 5

SocketPlugin extions
by Andreas Raab 27 Mar '06

27 Mar '06

Hi Guys - Attached the extensions for SocketPlugin which are based on David's code (except I've folded them directly into SocketPlugin). It's basically just two functions: a) Being able to mark the socket as "reusable", e.g., setting SO_REUSADDR/SO_REUSEPORT explicitly (the main reason for putting this into a separate primitive is because on different platforms subtly different actions need to be taken) b) Being able to bind() a socket explicitly to an interface/port. Cheers, - Andreas

2 3

Proposal: getSystemAttribute(1006) - build ID
by Andreas Raab 27 Mar '06

27 Mar '06

Hi Guys - Here is a short proposal for extending getSystemAttribute (see http://minnow.cc.gatech.edu/squeak/314): * 1006 - build id; any string that uniquely identifies a VM build, preferredly in UUID format ('aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeeeee') Is this okay with everybody? (if so, I'll update the above page to include it) Cheers, - Andreas

4 7

MPEG3Plugin problem
by Andreas Raab 26 Mar '06

26 Mar '06

Hi Guys - Looks like a declaration in MPEG3Plugin is screwed up: MPEG3Plugin>>primitiveMPEG3EndOfVideo: fileHandle stream: aNumber | file result | "int mpeg3_end_of_video(mpeg3_t *file, int stream)" self var: #file type: ''. translates to: EXPORT(sqInt) primitiveMPEG3EndOfVideo(void) { file; which (for the obvious reasons) doesn't work. Can somebody provide a quick fix? Thanks, - Andreas

3 9

Mac FFI prims plugin 1.6.0b3 pending (Croquet support)
by John M McIntosh 25 Mar '06

25 Mar '06

I'll note that the 3.8.11b4 mac carbon Universal VM uses a FFI plugin that is broken for Croquet. This seems to be a problem based on the FFI code used by the unix tree. I have migrated the project to use the older mac carbon FFI C code, and Andrea's windows FFI C code which then avoids the crash. These changes have been checked into the SVN tree. I'll post a 3.8.11b5 VM in a few hours or in a day or so since I'm still working on the OpenGL in a browser window support. Any earlier adopters can email me for a replacement plugin for testing. -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

3 3

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev March 2006