http://blog.gmane.org/gmane.comp.lang.ruby.core hourly 1 1901-01-01T00:00+00:00 http://gmane.org/img/gmane-25t.png http://gmane.org http://comments.gmane.org/gmane.comp.lang.ruby.core/19584 Hi list I was just having some real fun with the 191p2 and somehow I was sure that tail recursion would be optimized away, but alas.... I than searched the list for any information on that and did not find anything. Did I only dream or am I correct in assuming that Ruby1.9 will eventually optimize tail recursions? Thx in advance Cheers Robert Robert Dober 2008-12-03T21:47:20 http://comments.gmane.org/gmane.comp.lang.ruby.core/19563 Bug #816: Unexpected syntax error with: if true and not\n false http://redmine.ruby-lang.org/issues/show/816 Author: Joao Pedrosa Status: Open, Priority: Normal ruby -v so.rb ruby 1.8.7 (2008-11-11 patchlevel 73) [i686-linux] "yay" ruby1.9 -v so.rb ruby 1.9.1 (2008-12-01 revision 20437) [i686-linux] so.rb:2: syntax error, unexpected '\n' so.rb:3: warning: unused literal ignored so.rb:5: syntax error, unexpected keyword_end, expecting $end cat so.rb if true and not false p 'yay' end ---------------------------------------- http://redmine.ruby-lang.org Joao Pedrosa 2008-12-03T06:32:29 http://comments.gmane.org/gmane.comp.lang.ruby.core/19552 Bug #814: NoMethodError: undefined method `read_nonblock' for #<OpenSSL::SSL::SSLSocket:0x1a64f9a0> http://redmine.ruby-lang.org/issues/show/814 Author: Aaron Patterson Status: Open, Priority: Normal Category: ext My patch from #806 broke because OpenSSL::SSL::SSLSocket doesn't implement read_nonblock. I've attached a patch that implements read_nonblock on OpenSSL::SSL::SSLSocket. This is a naive patch. ---------------------------------------- http://redmine.ruby-lang.org Aaron Patterson 2008-12-03T04:54:48 http://comments.gmane.org/gmane.comp.lang.ruby.core/19548 irb(main):023:0> ->(c=1+2) {}.parameters => [[:opt, :c, 1]] Dave Thomas 2008-12-03T04:30:34 http://comments.gmane.org/gmane.comp.lang.ruby.core/19546 Merb uses autoload rather extensively. We have lately observed some disturbing behavior around concurrency. Effectively, because autoload removes the flag before loading the file, if two threads concurrently attempt to access an autoloaded constant, one of the threads can get a NameError. It's easy to reproduce this by adding a sleep to the file being loaded before the constant is defined, and then spinning up two threads that both try to use the constant. This is reproducible in MRI; it does not only happen with JRuby's true parallel execution. The worst part is that no user action can solve this problem: because autoloading is magic, it's impossible for the user to lock the resulting require. In effect, this makes autoloading completely useless for threaded environments. Is this intentional? Is there something we can do at the language level to ameliorate this problem? Yehuda Katz 2008-12-03T04:23:22 http://comments.gmane.org/gmane.comp.lang.ruby.core/19537 #encoding: utf-8 p " ∂og ".match(/\b./u) # => #<MatchData "o"> I was surprised that \b didn't find the boundary between the space and the Unicode ∂ character. Is that correct? Dave Dave Thomas 2008-12-02T23:46:31 http://comments.gmane.org/gmane.comp.lang.ruby.core/19525 Several people have played around with solutions that use ParseTree or RubyParser to extract the contents of a lambda and perform certain operations. For instance, Article.select {|a| a.title =~ /Some title/ } would generate "select * from articles where title ~ 'Some title'" This requires access to the ParseTree at runtime, which requires converting a proc to a ParseTree at runtime. I've been toying with the idea of an explicit language feature, like lisp quoting, that would explicitly pass the ParseTree to the method called. Example: Article.select(defer {|a| a.title =~ /Some title/ }) class Article def self.select(sexp) # sexp is: s(:defer, s(:lasgn, :a), s(:match3, s(:lit, /Some title/), s(:call, s(:lvar, :a), :title, s(:arglist)))) end end I understand that this would require a more explicit definition of the parse tree, but it seems like UnifiedRuby has provided a viable starting point for that work. Is this a terrible idea? Has it already been proposed and rejected innumerable times? Yehuda Katz 2008-12-02T17:36:18 http://comments.gmane.org/gmane.comp.lang.ruby.core/19524 Bug #811: Documentation Patch: Preventing XPath Injection attacks http://redmine.ruby-lang.org/issues/show/811 Author: Ken Bloom Status: Open, Priority: Low Here's a patch to rexml/xpath.rb which documents the variables parameter in REXML::XPath. I have previously posted this as ruby-core:16626, but it was ignored. Please apply it to ruby 1.8 and ruby 1.9, to encourage secure coding practices. --- xpath.rb.old2008-04-24 17:31:51.000000000 -0500 +++ xpath.rb2008-04-24 17:37:38.000000000 -0500 < at >< at > -15,10 +15,15 < at >< at > # node matching '*'. # namespaces:: # If supplied, a Hash which defines a namespace mapping. +# variables:: +# If supplied, a Hash which maps $variables in the query +# to values. This can be used to avoid XPath injection attacks +# or to automatically handle escaping string values. # # XPath.first( node ) # XPath.first( doc, "//b"} ) # XPath.first( node, "a/x:b", { "x"=>"http://doofus" } ) +# XPath.first( node, '/book/publisher/text()=$publisher', {}, {"publisher"=>"O'Reilly"}) def XPath::first element, path=nil, namespaces=nil, variables={} raise "The namespaces argument, if supplied, must be a hash object." unless namespaces.nil? or namespaces.kind_of?(Hash) raise "The variables argument, if supplied, must be a hash object." unless variables.kind_of?(Hash) < at >< at > -38,10 +43,16 < at >< at > # The xpath to search for. If not supplied or nil, defaults to '*' # namespaces:: # If supplied, a Hash which defines a namespace mapping +# variables:: +# If supplied, a Hash which maps $variables in the query +# to values. This can be used to avoid XPath injection attacks +# or to automatically handle escaping string values. # # XPath.each( node ) { |el| ... } # XPath.each( node, '/*[< at >attr='v']' ) { |el| ... } # XPath.each( node, 'ancestor::x' ) { |el| ... } +# XPath.each( node, '/book/publisher/text()=$publisher', {}, {"publisher"=>"O'Reilly"}) \ +# {|el| ... } def XPath::each element, path=nil, namespaces=nil, variables={}, &block raise "The namespaces argument, if supplied, must be a hash object." unless namespaces.nil? or namespaces.kind_of?(Hash) raise "The variables argument, if supplied, must be a hash object." unless variables.kind_of?(Hash) ---------------------------------------- http://redmine.ruby-lang.org Ken Bloom 2008-12-02T15:29:47 http://comments.gmane.org/gmane.comp.lang.ruby.core/19523 Bug #808: IO (StringIO) documentation corrections. http://redmine.ruby-lang.org/issues/show/808 Author: Emiel van de Laar Status: Open, Priority: Normal Category: DOC, Target version: 1.9.1 Hello, I went through io.c and stringio.c and corrected the documentation for a couple of methods (new way of handling strings). Patch is attached. Thanks! ---------------------------------------- http://redmine.ruby-lang.org Emiel van de Laar 2008-12-02T13:00:10 http://comments.gmane.org/gmane.comp.lang.ruby.core/19512 Hi, Juhapekka Tolvanen has informed me that unicode.org has a list of Unicode aware languages, which does not include Ruby. http://www.unicode.org/resources/programs.html He suggested to contact Unicode.org, but it might be a task suitable for a more fluent English speaker than me. Could any one contact them in place of me? Probably contacting place would be: http://www.unicode.org/reporting.html matz. Yukihiro Matsumoto 2008-12-02T06:40:47 http://comments.gmane.org/gmane.comp.lang.ruby.core/19502 Bug #806: Net::Protocol#rbuf_fill is slow http://redmine.ruby-lang.org/issues/show/806 Author: Aaron Patterson Status: Open, Priority: Normal Category: lib Net::Protocol#rbuf_fill is slow. We can take advantage of io.read_nonblock and IO.select to deal with timeouts. I've attached a patch. ---------------------------------------- http://redmine.ruby-lang.org Aaron Patterson 2008-12-01T18:22:32 http://comments.gmane.org/gmane.comp.lang.ruby.core/19501 Hi everyone, While working on some specs, we found out that autoload (Kernel#autoload and Module#autoload) internally call rb_require() directly to pull the file in. This of course means that if Kernel#require has been changed (for instance by rubygems) that the changed logic won't be run. My question is if it's intentional that autoload bypasses Kernel#require. If not, autoload should probably use rb_funcall() to keep everything consistent. This obviously also effects Rubinius, as our autoload just calls Kernel#require directly. Thanks! - Evan Phoenix Evan Phoenix 2008-12-01T17:54:36 http://comments.gmane.org/gmane.comp.lang.ruby.core/19500 Bug #805: Ruby 1.9.1 preview 2 : build failure on OpenSolaris http://redmine.ruby-lang.org/issues/show/805 Author: Dae San Hwang Status: Open, Priority: High I got the following error while building 1.9.1 preview 2 on OpenSolaris 5.11 snv_98. (1.9.1 preview 1 was built fine on OpenSolaris.) It was configured with the following: ./configure --build=x86_64-sun-solaris10 --prefix=/usr/local CFLAGS="-m64" LDFLAGS="-m64" -- compiling curses make[1]: Entering directory `/usr/local/src/ruby-1.9.1-preview2/ext/curses' gcc -I. -I../../.ext/include/x86_64-solaris10 -I../.././include -I../.././ext/curses -DRUBY_EXTCONF_H=\"extconf.h\" -fPIC -m64 -O2 -g -Wall -Wno-parentheses -o curses.o -c curses.c curses.c:419:18: macro "ISPRINT" requires 2 arguments, but only 1 given curses.c: In function `curses_getch': curses.c:419: error: `ISPRINT' undeclared (first use in this function) curses.c:419: error: (Each undeclared identifier is reported only once curses.c:419: error: for each function it appears in.) curses.c:1113:18: macro "ISPRINT" requires 2 arguments, but only 1 given curses.c: In function `window_getch': curses.c:1113: error: `ISPRINT' undeclared (first use in this function) make[1]: *** [curses.o] Error 1 make[1]: Leaving directory `/usr/local/src/ruby-1.9.1-preview2/ext/curses' make: *** [exts] Error 1 ---------------------------------------- http://redmine.ruby-lang.org Dae San Hwang 2008-12-01T16:50:41 http://comments.gmane.org/gmane.comp.lang.ruby.core/19499 Bug #804: Ruby 1.9.1 preview 2 : make test failure http://redmine.ruby-lang.org/issues/show/804 Author: Philippe Lucas Status: Open, Priority: Normal Version built on Fedora core 2 x686 with gcc 4.0.2 in a sub directory (UNIX) of the main ruby source with ./configure --prefix=/h/phil/DEV/RUBY/ruby-191pv2 --enable-shared --enable-install-doc --enable-pthread --with-mantype=man ./ruby -v => ruby 1.9.1 (2008-12-01 revision 20438) [i686-linux] Note that : ruby -v => ruby 1.8.7 (2008-05-31 patchlevel 0) [i686-linux] In the announce message of this release, i read : == Improvements * You can "make test" without installing ruby. So make test gives this error : test_flip.rb /h/phil/DEV/RUBY/ruby-1.9.1-preview2/UNIX/ruby: error while loading shared libraries: libruby.so.1.9: cannot open shared object file: No such file or directory which is not very suprising !. Regards. ---------------------------------------- http://redmine.ruby-lang.org Philippe Lucas 2008-12-01T15:44:35 http://comments.gmane.org/gmane.comp.lang.ruby.core/19498 Ruby 1.9.1 Preview 2 has been released. This is a preview release of Ruby 1.9, which will be the first stable version of Ruby 1.9 series. If you encounter a bug or a problem, please let us know it via the official issue tracking system (http://redmine.ruby-lang.org ). == Location * ftp://ftp.ruby-lang.org/pub/ruby/1.9/ruby-1.9.1-preview2.tar.bz2 SIZE: 6148131 bytes MD5: 62126475998ede5318c1bc82c40d5f48 SHA256: 2c419dc325c6a75fb7b961496c0dd54f2729e6e01730589c4fb06e34ddd7a7cc * ftp://ftp.ruby-lang.org/pub/ruby/1.9/ruby-1.9.1-preview2.tar.gz SIZE: 7375483 bytes MD5: 7699b9e54c53b16640d40a213588e704 SHA256: dd737d26212d68c26ecb4d6e2ed180a0d7d5e27fb7bd2c819935494daa1cf50b * ftp://ftp.ruby-lang.org/pub/ruby/1.9/ruby-1.9.1-preview2.zip SIZE: 8612618 bytes MD5: d3ae200280b75801bfc5c090d3592577 SHA256: 96b6a2745b64f03ffbd5da97f93a2676365e3900cb8f6bf820f47739ef8d3fa9 == Improvements * supports 21 new character encodings. * works better with RubyGems. * handles SEGV better when system stack overflows. * fixed a bit of regexp imcompatiblity to Perl5 and old Ruby. * many bugs in Time#strftime and Date#strftime fixed. * process termination got faster. * allows long style options in RUBYOPT. * You can "make test" without installing ruby. * You can derive a class from Fiber correctly. * many bugs fixed. == Release schedule Current Status: * Language features are effectively "frozen". * But we could not finish multilingualization tasks for standard libraries. For example, irb does not recognize magic comments. * So changes to some of standard libraries continue. * irb * erb * cgi * ... Schedule: * 25 Dec, 2008: Ruby 1.9.1 release candidate * 25 Jan, 2009: Ruby 1.9.1 Thanks Yugui (Yuki Sonoda 2008-12-01T13:25:50 http://comments.gmane.org/gmane.comp.lang.ruby.core/19487 Hi, Windows Vista (Home Premium) MS VC++ 9 (cl 15) Ruby 1.8.6-p114 and Ruby 1.8.6-p237 eval.c ... eval.c(9839) : fatal error C1189: #error : unsupported platform NMAKE : fatal error U1077: '"C:\Program Files (x86)\Microsoft Visual Studio 9.0\VC\BIN\amd64\cl.EXE"' : return code '0x2 ' Stop. Where the relevant code is: #if !defined SAVE_WIN32_EXCEPTION_LIST && !defined _WIN32_WCE # error unsupported platform #endif #endif Regards, Dan Daniel Berger 2008-11-30T13:42:36 http://comments.gmane.org/gmane.comp.lang.ruby.core/19472 I noticed that the following reg exp causes syntax error in 1.9.1 (I use tip of the branch): /\n(\.\/)?(.*spec\.rb):[\d]+:\Z?/ To reproduce, just fire up irb of 1.9.1 and try to assign this literal to some variable. The problem is with ? quantifier following \Z. I could not find any information about this restriction on the web and have no Mastering regular expressions book at hand. Could someone shed some light on whether it is 1.9.1 engine restriction, general regular expressions restriction or possible 1.9.1 bug? Just in case: this regexp comes from autospec in RSpec. Thank you, MK Michael Klishin 2008-11-28T23:14:36 http://comments.gmane.org/gmane.comp.lang.ruby.core/19469 Bug #800: Problem with Array class http://redmine.ruby-lang.org/issues/show/800 Author: Francois Proulx Status: Open, Priority: High If you test this simple program, you will see why array should be dup on assignment ... a=1 b=1 c=[a,b] def foo(x) x[0] +=1 x[1] +=1 end d = c e = d f = e foo(f) p c[0],c[1] ---------------------------------------- http://redmine.ruby-lang.org Francois Proulx 2008-11-28T20:34:20 http://comments.gmane.org/gmane.comp.lang.ruby.core/19442 Bug #799: cross compile 1.9.1-preview1 is not possible http://redmine.ruby-lang.org/issues/show/799 Author: Luis Lavena Status: Open, Priority: Normal Target version: 1.9.1 Hello everybody. I tried to cross compile 1.9.1-preview1 under ubuntu. It succeeded with 1.8.6-p287 and p114 but failed with 1.9. Used the following compile flags: <pre><code>--host=i586-mingw32msvc --target=i386-mingw32 --build=i686-linux --enable-shared</code></pre> This is the snipped of the output <pre><code>make[1]: Leaving directory `/home/luislavena/sources/cross-ruby/builds/ruby-1.9.1-preview1' make -f enc.mk RUBY="ruby -I/home/luislavena/sources/cross-ruby/builds/ruby-1.9.1-preview1 -rfake " MINIRUBY="ruby -I/home/luislavena/sources/cross-ruby/builds/ruby-1.9.1-preview1 -rfake " encs make[1]: Entering directory `/home/luislavena/sources/cross-ruby/builds/ruby-1.9.1-preview1' mkdir -p .ext/i386-mingw32/enc .ext/i386-mingw32/enc/trans enc enc/trans make[1]: Leaving directory `/home/luislavena/sources/cross-ruby/builds/ruby-1.9.1-preview1' /home/luislavena/sources/cross-ruby/sources/ruby-1.9.1-preview1/ext/extmk.rb:338: undefined method `<<' for nil:NilClass (NoMethodError) make: *** [exts] Error 1 </code></pre> Attached is the complete output of my rake script. ---------------------------------------- http://redmine.ruby-lang.org Luis Lavena 2008-11-26T21:57:39 http://comments.gmane.org/gmane.comp.lang.ruby.core/19440 I'm probably missing something trivial, but given the following Ruby code: class Myclass < at >< at >myvar = 1 def self.myvar return < at >< at >myvar end end I can access < at >< at >myvar from Ruby with Myclass::myvar. Great, but now I need to access that same variable through a C extension. I think I can get it by doing: VALUE myvar = rb_eval_string("Myclass::myvar"); but surely there's a better way? I'm thinking something like rb_cvar_get() but that requires that I pass in a VALUE and all I have is a string at this point, "Myclass". Christopher Thompson 2008-11-26T19:49:32 http://comments.gmane.org/gmane.comp.lang.ruby.core/19436 Hello ruby-core, Today I was playing around with manipulating strings containing binary data, i.e. "\xaa\xab\xac\xad\xae" and using the new String methods available in Ruby 1.9. The exercise I was trying out was to extract out a range of bytes as Fixnums. Kind of like String#bytes but I was only interested in a subarray. Like so: "\xaa\xab\xac\xad\xae".bytes.to_a[1,3] # => [171, 172, 173] This works but operates on the entire data set which I imagine is fairly expensive... So I chopped it up before hand like so: "\xaa\xab\xac\xad\xae"[1,3].bytes.to_a => [171, 172, 173] Much better. This works fine when using the ASCII-8BIT encoding. As soon as you use something like UTF-8 it fails because the [] method now works on characters instead of bytes. data = "\xc3\xa9\xc3\xa9" # => "\xC3\xA9\xC3\xA9" data.force_encoding("utf-8") # => "" data[0,2].bytes.to_a # => [195, 169, 195, 169] Here I get four bytes instead of the first two which I wanted. data.bytes.to_a[0,2] # => [195, 169] So having said that I must ensure that the binary data I am working with is encoded using ASCII-8BIT. This is off course completely reasonable and recommended. Anyway, my real comment is that it might be nice to have String#getbyte or String#bytes be able to get a subset of bytes from the string. For example: "abcde".bytes(1,3).to_a # => [98, 99, 100] "abcde".bytes(1..3).to_a # => [98, 99, 100] "abcde".bytes(-2,2).to_a # => [100, 101] "abcde".getbyte(1,3) # => [98, 99, 100] "abcde".getbyte(1..3) # => [98, 99, 100] "abcde".getbyte(-2,2) # => [100, 101] String#getbyte is singular as opposed to plurar which doesn't sit well with me. Thanks for reading! - Emiel van de Laar lar as opposed to Emiel van de Laar 2008-11-26T16:07:08 Search the mailing list at Gmane query http://search.gmane.org/?group=$group=gmane.comp.lang.ruby.core