January 2011 - MacRuby-devel - lists.macosforge.org

Dispatch gem - proxy object
by Alan Skipp 29 Jan '11

29 Jan '11

Hi everyone, After the recent discussion about gcd and thread safety I took a look at the Dispatch gem (proxy object in particular). I though I'd try and improve efficiency, as all method calls are currently directed through method_missing to the delegate object. One attempt I made was to dynamically define methods in method_missing so that subsequent calls would be more efficient. I realised the naivety of this approach when one queue was busily defining the method whilst other queues were calling method_missing - much pain ensued. The next approach might be more promising: remove method_missing and instead define all methods of the delegate object in initialize. It appears to work most of the time, but unfortunately crashes are not infrequent. Is there anything obviously wrong about the initialize method below? Cheers, Alan ------ require 'delegate' module Dispatch class Proxy < SimpleDelegator attr_accessor :__group__, :__queue__, :__sync__ def initialize(delegate, group=Group.new, queue=Dispatch::Queue.concurrent) super(delegate) @__serial__ = Dispatch::Queue.for(self) @__group__ = group @__queue__ = queue @__retval__ = nil delegate.public_methods.each do |meth| (class << self; self; end).class_eval do define_method meth do |*args, &block| if block.nil? then @__serial__.sync { @__retval__ = __getobj__.__send__(meth, *args) } return @__retval__ end queue = @__queue__ # copy in case it changes while in flight @__serial__.async(@__group__) do retval = __getobj__.__send__(meth, *args) queue.async(@__group__) { block.call(retval) } end end end end end end

1 1

Thread safety in apply example?
by Charles Oliver Nutter 29 Jan '11

29 Jan '11

I'm curious about this example in Queue#apply's rdoc: * gcdq = Dispatch::Queue.new('doc') * @result = [] * gcdq.apply(5) {|i| @result[i] = i*i } * p @result #=> [0, 1, 4, 9, 16, 25] apply is said to issue the jobs in parallel, so this would be making concurrent updates to the @result array. Are simple arrays in MacRuby thread-safe? - Charlie

8 22

Can't build the MacRuby latest.
by Watson 28 Jan '11

28 Jan '11

Hi, Now, I'm rebuilding all of MacRuby latest. But, it seems that does not finish forever with following build log. ----snip---- ./miniruby -I. -I./lib bin/rubyc --internal --arch x86_64 -C "ext/ripper/lib/ripper/filter.rb" -o "ext/ripper/lib/ripper/filter.rbo" ./miniruby -I. -I./lib bin/rubyc --internal --arch x86_64 -C "ext/ripper/lib/ripper/lexer.rb" -o "ext/ripper/lib/ripper/lexer.rbo" ./miniruby -I. -I./lib bin/rubyc --internal --arch x86_64 -C "ext/ripper/lib/ripper/sexp.rb" -o "ext/ripper/lib/ripper/sexp.rbo" ./miniruby -I. -I./lib bin/rubyc --internal --arch x86_64 -C "ext/ripper/lib/ripper.rb" -o "ext/ripper/lib/ripper.rbo" cd ext/bigdecimal && ../../miniruby -I../.. -I../../lib -r rbconfig -e "RbConfig::CONFIG['libdir'] = '../..'; require './extconf.rb'" checking for labs() in stdlib.h... yes checking for llabs() in stdlib.h... yes creating Makefile ------------- Does it occur only in my environment? Thank you.

3 4

Strange NSDate behavior building 32 bit v 64 bit
by Richard Sepulveda 28 Jan '11

28 Jan '11

I am seeing some strange behavior with NSDate, here is my basic code in a new MacRuby project. I am running on a MacBook 2009 model, Mac OSX 10.6.6 I am using the latest XCode and the MacRuby Framework 0.7 in XCode in rbmain.rb: NSLog("In rb_main.rb") puts "NSDate = " + NSDate.date.to_s NSLog( "NSDate = " + NSDate.date.to_s) CCode.printNSDate in CCode.m: @implementation CCode + (void) printNSDate { NSLog(@"In printNSDate, NSDate = %@", NSDate.date); } @end When compiled: 64-bit Universal, the output is 2011-01-28 05:54:16.972 TestRuby[90066:a0f] In rb_main.rb NSDate = 2011-01-28 05:54:16 -0600 2011-01-28 05:54:16.976 TestRuby[90066:a0f] NSDate = 2011-01-28 05:54:16 -0600 2011-01-28 05:54:16.979 TestRuby[90066:a0f] In printNSDate, NSDate = 2011-01-28 05:54:16 -0600 When compiled: 32-bit Universal, the output is 2011-01-28 05:51:23.941 TestRuby[89998:a0f] In rb_main.rb NSDate = 2011-01-28 05:44:00 -0600 2011-01-28 05:51:23.958 TestRuby[89998:a0f] NSDate = 2011-01-28 05:44:00 -0600 2011-01-28 05:51:23.961 TestRuby[89998:a0f] In printNSDate, NSDate = 2011-01-28 05:51:23 -0600 2011-01-28 05:51:24.565 TestRuby[89998:a0f] Error loading /Library/InputManagers/Smart Crash Reports/Smart Crash Reports.bundle/Contents/MacOS/Smart Crash Reports: dlopen(/Library/InputManagers/Smart Crash Reports/Smart Crash Reports.bundle/Contents/MacOS/Smart Crash Reports, 265): no suitable image found. Did find: /Library/InputManagers/Smart Crash Reports/Smart Crash Reports.bundle/Contents/MacOS/Smart Crash Reports: GC capability mismatch 2011-01-28 05:51:25.409 TestRuby[89998:a0f] Error loading /Library/InputManagers/SIMBL/SIMBL.bundle/Contents/MacOS/SIMBL: dlopen(/Library/InputManagers/SIMBL/SIMBL.bundle/Contents/MacOS/SIMBL, 265): no suitable image found. Did find: /Library/InputManagers/SIMBL/SIMBL.bundle/Contents/MacOS/SIMBL: GC capability mismatch The Obj-C routine prints the correct time. 5:51:23 in both cases. For 32 bit, the Ruby code prints some time in the past that never changes. For 64 bit, the Ruby code prints the correct time. Any idea's what could be going on? Thanks!

1 0

0.9 release
by Laurent Sansonetti 28 Jan '11

28 Jan '11

Hi guys, I think it's time to release trunk as 0.9. I expect to do it next week, if possible. I tagged the following tickets as blockers: http://www.macruby.org/trac/ticket/1121 http://www.macruby.org/trac/ticket/1128 If you think another ticket should also be a 0.9 blocker let me know, as I may have overlooked the list :) We are looking for functional or performance regressions. If you're doing a Cocoa app, please try a recent nightly build and let me know if anything is broken, as we must not ship any regression at this point. Thanks :) Laurent

3 4

Bizarre behaviour with an instance variable
by Martin Hawkins 27 Jan '11

27 Jan '11

I find the following behaviour a little bizarre: At the macirb prompt (0.8) fileManager = NSFileManager.defaultManager @theProjectRoot = '/Users/martin/work/macruby/lots_of_files' => "/Users/martin/work/macruby/lots_of_files" >> filenames = fileManager.contentsOfDirectoryAtPath(@theProjectRoot, error:nil).map {|f| theProjectRoot.stringByAppendingPathComponent(f) } NameError: undefined local variable or method `theProjectRoot' for main:TopLevel If I create a local variable with the name theProjectRoot, then the statement runs correctly. According to the documentation, NSFileManager.defaultManager.contentsOfDirectoryAtPath requires a string; the scope of the variable should be irrelevant, shouldn't it?

3 4

Test problem
by Martin Hawkins 27 Jan '11

27 Jan '11

I have a Classes/Helpers/Constants.rb module . Several classes use it, so the following is common in my code:e.g # File: Record.rb require 'Constants' include Constants class Record end I have a test for Record.rb that worked prior to to my introduction of Constants: require 'test/unit' class RecordTest < Test::Unit::TestCase etc end Since introducing the Constants module, I get 'run_suite.rb:3:in 'block': no such file to load -- Constants -- (LoadError) I need to tell XCode where to find the Constants.rb file but I don't know how. Can somebody please point me in the right direction?

1 0

Setting properties on a dereferenced NSRangePointer has no effect
by Caio Chassot 26 Jan '11

26 Jan '11

Hi all, NSFormatter defines a method with the following selector: isPartialStringValid:proposedSelectedRange:originalString:originalSelectedRange:errorDescription: This is its signature: - (BOOL)isPartialStringValid:(NSString **)partialStringPtr proposedSelectedRange:(NSRangePointer)proposedSelRangePtr originalString:(NSString *)origString originalSelectedRange:(NSRange)origSelRange errorDescription:(NSString **)error The relevant bit here is the proposedSelRangePtr argument, which is a NSRangePointer. When subclassing NSFormatter in ruby, I define the method as: def isPartialStringValid(partialStringPtr, proposedSelectedRange:proposedSelRangePtr, originalString:origString, originalSelectedRange:origSelRange, errorDescription:error) When implementing the Cocoa Programming exercises in MacRuby, I ran across the following situation: At some point in that method definition, in the original code, Aaron sets the range properties directly: proposedSelRangePtr->location = [*partialStringPtr length]; proposedSelRangePtr->length = [match length] - proposedSelRangePtr->location; Initially, I did similar in MacRuby proposedSelRangePtr[0].location = partialStringPtr[0].length proposedSelRangePtr[0].length = match.length - proposedSelRangePtr[0].location This has no effect. It does not raise, but the values go unchanged. I can NSLog the range values before and after and they're the same. My final solution was to assign a new range to that location: proposedSelRangePtr[0] = NSRange.new(partialStringPtr[0].length, match.length - partialStringPtr[0].length) But I wonder why setting the range properties had no effect. MacRuby bug?

2 2

Re: [MacRuby-devel] [macruby-changes] [5186] MacRuby/trunk/hash.h
by Laurent Sansonetti 26 Jan '11

26 Jan '11

Hi Watson, This isn't good, rhash_modify() must be very fast, so calling OBJ_FROZEN and OBJ_UNTRUSTED is not good there. Can we look up the mask flag as before? Laurent On Jan 21, 2011, at 7:51 AM, source_changes(a)macosforge.org wrote: > Revision > 5186 > Author > watson1978(a)gmail.com > Date > 2011-01-21 07:51:01 -0800 (Fri, 21 Jan 2011) > Log Message > > More method of Hash will throw a SecurityError when $SAFE is 4. > > Test Script: > {{{ > h = {} > $SAFE = 4 > h['a'] = 1.0 > }}} > Modified Paths > > MacRuby/trunk/hash.h > Diff > > Modified: MacRuby/trunk/hash.h (5185 => 5186) > > --- MacRuby/trunk/hash.h 2011-01-21 02:20:08 UTC (rev 5185) > +++ MacRuby/trunk/hash.h 2011-01-21 15:51:01 UTC (rev 5186) > @@ -41,14 +41,11 @@ > static inline void > rhash_modify(VALUE hash) > { > - const long mask = RBASIC(hash)->flags; > - if ((mask & FL_FREEZE) == FL_FREEZE) { > - rb_raise(rb_eRuntimeError, "can't modify frozen/immutable hash"); > + if (OBJ_FROZEN(hash)) { > + rb_error_frozen("hash"); > } > - if ((mask & FL_TAINT) == FL_TAINT) { > - if (rb_safe_level() >= 4) { > - rb_raise(rb_eSecurityError, "Insecure: can't modify hash"); > - } > + if (!OBJ_UNTRUSTED(hash) && rb_safe_level() >= 4) { > + rb_raise(rb_eSecurityError, "Insecure: can't modify hash"); > } > } > > _______________________________________________ > macruby-changes mailing list > macruby-changes(a)lists.macosforge.org > http://lists.macosforge.org/mailman/listinfo.cgi/macruby-changes

2 2

Extending MacRuby to platforms other than OSX?
by Adam Strzelecki 25 Jan '11

25 Jan '11

Hi, First of all I think MacRuby is amazing project, and one and only Ruby runtime/compiler (macrubyc) that is able to produce standalone self-contained binaries. Since I have used Ruby heavily to produce server-side applications rather than Cocoa based application and I am iPhone developer as well, I wish to ask if and what are the plans for extending MacRuby to other platforms like: (1) iOS - here I know the basic problem is lack of Obj-C GC. I've seen some post you're working on it, any clues about that? (2) any UNIX (POSIX) platforms (Linux servers) - most of tech MacRuby relays on is OpenSource including LLVM, Obj-C runtime... most except Foundation framework is closed-source and exists only on OSX, however there's CFLite and there were couple of tries to port Foundation to generic UNIX (POSIX), one of them is PureFoundation. Do you guys considered stripped down MacRuby version for any POSIX OS where Array, String and some core classes use some alternative implementation of Foundation or Foundation-free implementation? Regards, -- Adam Strzelecki | nanoant.com | tel. +48 605 446 492

5 6