1651-1905 subjects 2057-2339

[PATCH] exit inside test/unit
1836 [nobu.nokada@] Now test/unit terminates whole process at the end of tests,
1845 [nathaniel@ta] Your patch looks good, but I think we need to go further...
+ 1860 [ryand-ruby@z] Ditto.
| 1861 [nathaniel@ta] The at_exit is currently one line of code; the content has recently
+ 1863 [nobu.nokada@] This would be necessary, though I missed.
  1865 [nathaniel@ta] I'd prefer something more explicit than 'trb', perhaps 'testruby',
  1870 [nobu.nokada@] The last sounds good to me.
  1872 [nathaniel@ta] Excellent.

DRb tests hang on OS X 10.3.1
1843 [nathaniel@ta] I haven't yet been able to test this on another platform to see if it
1848 [s-urabe@pa .] Curious, I cannot confirm this.
1850 [nathaniel@ta] Strange... are you also running 10.3.1?
1864 [s-urabe@pa .] Yes. I'm using build 7C107.
1888 [nathaniel@ta] Running with -d revealed that the problem was due to having the Mac OS

Ruby2 RCR (was Re: Constants, class variables and the cbase field)
1847 [matz@ru y- a] I'm sick of proposals that sounds like "hey, matz, I'm a Ruby newbie;
+ 1849 [george.marro] Matz -- sincere apologies for exasperating you. It wasn't ever my intention
| 1851 [matz@ru y- a] I was afraid you said so.  It was not my intention.  You didn't preach
+ 1857 [transami@ru ] I know I'm probably the last person on earth you, or anyone, wants to hear
  + 1859 [mwilson13@co] Why is this discussion happening on ruby-core and not on ruby-talk?
  + 1862 [matz@ru y- a] inferring duck typing, and method combination with redefining and
    1866 [transami@ru ] Thank you, matz. That is very considerate of you to say. Not to worry. I don't
    1867 [matz@ru y- a] I put that in the ToDo file, just because CommonLisp has that
    1868 [transami@ru ] Okay. Tha't fine. It would have been nice to have known about before hand. As
    1869 [transami@ru ] s/tired/sick/

WEBrick bug & fix
1853 [nathaniel@ta] I have a production WEBrick server whose common usage includes clients
1882 [gotoyuzo@no ] Thanks. I understood the situation.

The Ancient Prophecies and the Day of Death (was Re: New syck bug)
1856 [ruby-core@wh] We'll be fine.  Like I've said before: As long as you've all got about a
1858 [rich@in oe h] Thanks _why, your eschatological interpretations comfort me.  I couldn't

[PATCH] Old Yacc limitation
1873 [sunshine@su ] Some old versions of Yacc (such as the NextStep Yacc) severely limit
1874 [matz@ru y- a] This will be merged.  Thank you.

PATCH: ftools.rb, to aid RDoc processing
1878 [gsinclair@so] I don't commit code patches without permission, so can someone OK
1879 [dave@pr gp o] Actually, I'm about to change this so that RDoc will implicitly declare

[PATCH] OpenSSL::X509::Certificate#inspect bug
1880 [guillaume.pi] /************************************************************************/
1881 [guillaume.pi] <bug>
1883 [gotoyuzo@no ] That was fixed. Thank you.

[PATCH] multiple exceptions for assert_raises
1884 [nobu.nokada@] Exceptions caused by system call often varies depending on
+ 1885 [matz@ru y- a] By the way "assert_raises" looks against the "plain form noun" rule of
| 1887 [nathaniel@ta] Hmmm... I wasn't aware of the rule. Is it documented somewhere?
| 1889 [matz@ru y- a] No.  I mentioned the rule on the ruby-talk list several times though.
| + 1890 [nathaniel@ta] Guess I missed it. Oh well. If I wasn't so lazy myself I'd suggest that
| | 1894 [matz@ru y- a] Should have been written somewhere.  Maybe in the FAQ.
| + 1892 [dblack@wo bl] Will is_a? become are_a?
|   1893 [matz@ru y- a] "is_a?" will _not_ be removed, since it's so widely used among OOP,
+ 1886 [nathaniel@ta] Looks good. Please commit.
  1891 [nathaniel@ta] I should've brought this up before, Nobu, but can you add tests for the
  1895 [nobu.nokada@] Commited, check it out, plz.
  1896 [nathaniel@ta] Hmmm... I don't see any changes to test_assertions.rb. Did you forget
  1897 [nobu.nokada@] Sorry, I missed to commit it.  Now committed with
  1898 [nathaniel@ta] Looks great, Nobu! Thanks!

1899 [nathaniel@ta] progress
1900 [djberge@qw s] Assuming things are consistent with regards to sockets between Perl and

Test::Unit problem
1901 [ser@ge ma e-] I'm having a curious problem: I keep getting the same error trying to run unit tests from the CVS distribution of Ruby.
+ 1904 [decoux@mo lo] Well, I don't remember exactly the problem but try to make *first*
+ 1906 [nobu.nokada@] I guess this "require" line is no longer needed.  This would
  1907 [ser@ge ma e-] Thanks, that was it.
  1911 [nobu.nokada@] The problem is caused by cyclic requiring between
  1912 [nathaniel@ta] This looks great, Nobu. Please commit.

css2.rb RDoc template error
1902 [transami@ru ] I love the look of the new css2.rb template in RDoc. Well, that is if it
1903 [dave@pr gp o] it seems to work here with the CVS version of RDoc from ruby-lang. If

unknown node type error 0 [Was: Test::Unit problem]
1908 [ser@ge ma e-] Actually, ignore that.  Even loading that file breaks the tests, even if you
+ 1909 [decoux@mo lo] This means that the node was gc'ed
+ 1910 [nobu.nokada@] It may be related to this bug, check your eval.c version.

1.8.1 release schedule?
1913 [nathaniel@ta] What is the planned release schedule for 1.8.1? Obviously it has been
1915 [matz@ru y- a] At most before Christmas, I hope.

-Wall warnings from 1.8.1 p3
1914 [djberge@qw s] Here are some potentially significant warnings from 1.8.1 p3
1919 [nobu.nokada@] Does following patch clean these warnings?
1920 [djberge@qw s] file.c:329: warning: unsigned int format, different type arg (arg 3)
+ 1921 [sunshine@su ] AC_CHECK_HEADERS([... ieeefp.h ...])
+ 1922 [nobu.nokada@] Replace %o with %lo.
  + 1923 [matz@ru y- a] No, since curses_attroff() etc. are used without USE_COLOR, #ifdef
  | 1924 [nobu.nokada@] curses_attroff() uses window_attroff(), which is not surrounded
  + 1925 [djberge@qw s] That fixes it.  Thanks.
    + 1926 [ehughes@bl e] it's not so much a "quirk" as an attempt to point out that you look like
    + 1930 [nobu.nokada@] In ruby.h, alloca.h was ignored for GCC.
      1931 [matz@ru y- a] Commit them.

Errors building 1.8.1p3
1916 [nathaniel@ta] I'm trying to build 1.8.1p3 on a Debian stable system (with some
1917 [nathaniel@ta] Oops... I just discovered that the error is something that occurred due
1918 [matz@ru y- a] Warnings are caused by conflicts in files asm/sigcontext.h and

Problem with REXML in Ruby CVS:
1927 [austin@ha os] I'm trying to check out the latest Ruby CVS on Windows, and I'm getting the
+ 1928 [a.bokovoy@sa] File systems supported in Windows are caseless -- contrary to most of
| 1951 [ser@ge ma e-] This has been fixed in CVS.
+ 1929 [austin@ha os] I forgot to mention that this is a fresh checkout. The problem is in the

--enable-pthread broken?
1932 [nathaniel@ta] [ruby-talk: 87759] and the surrounding thread seem to indicate that
1933 [matz@ru y- a] How often do you have crash?  I had two out of many tries.
1934 [nathaniel@ta] Every time I've run the test programs with --enable-pthread, one or the
1935 [decoux@mo lo] Well, look where it crash.
1937 [nobu.nokada@] Then, it'd be simple.
+ 1938 [nathaniel@ta] Maybe for you... but for me, I'm very grateful for Ruby super-heroes :-)
| 1939 [matz@ru y- a] That indicates me that I have to fight with [ruby-talk:87519] anyway.
| + 1941 [matz@ru y- a] Nathaniel, let me ask you few more questions before start fighting.
| | 1943 [nathaniel@ta] I can't say. --enable-pthread caused the segfault, so I don't think the
| | 1953 [matz@ru y- a] LinuxThread has something bad with signal handling, I guess.  This
| | 1959 [decoux@mo lo] If I've well understood it receive a signal when it's in setcontext().
| | 1961 [matz@ru y- a] Hm.
| | 1962 [decoux@mo lo] In rb_thread_restore_context() there is
| | 1963 [matz@ru y- a] I thought context switch would not happen during restore_context,
| | 1964 [decoux@mo lo] ruby has a new thread thread_timer() where it can call nanosleep() when
| | 1965 [matz@ru y- a] Sorry for being stupid, but rb_trap_immediate is turned off in the
| | + 1966 [decoux@mo lo] If I'm right this is the interaction between pthreads (POSIX) and rthreads
| | + 1967 [decoux@mo lo] here the case (svg can have the problem very quickly)
| | | 1970 [matz@ru y- a] Hmm, I cannot reproduce the blocking.  Maybe my machine is too slow.
| | | 1971 [decoux@mo lo] Personnaly I suspect that svg is stupid, this is why it block :-)
| | + 1968 [decoux@mo lo] Another example (yesterday it was easier to see that that was io_fflush()
| | + 1969 [nobu.nokada@] Isn't there a possibility that main pthread enters a critical
| |   1972 [matz@ru y- a] This fix seems reasonable, and Guy confirmed it changes the
| |   1985 [nathaniel@ta] OK, I've updated to the most recent CVS (I think all the patches have
| |   1987 [nobu.nokada@] Since the linux pthreads is implemented with CLONE(2), all
| |   1988 [nathaniel@ta] As long as it's expected, I'm fine with it. Thanks for the explanation.
| + 1942 [nathaniel@ta] Unfortunately it does look that way. pthreads turned out to be a rabbit
+ 1944 [decoux@mo lo] Seems to work with me.
  1947 [matz@ru y- a] Nobu, can you commit?

Can't define +@ for Symbol (plus ruby install problem)
1936 [transami@ru ] I wanted to see if the +@ problem was fixed in 1.8.1 preview 3 but when I do
1940 [matz@ru y- a] Unary plus is used as an identity operator.  Do you really want to use
1945 [transami@ru ] I'm putting together a small type library (maybe called lightype, not sure
1948 [matz@ru y- a] Hmm, I like
+ 1949 [s-urabe@pa .] What if I go crazy and want to define a method named +r ?
| 1952 [matz@ru y- a] You need to specify by "++r", "-+r", probably.
| 1954 [s-urabe@pa .] I see.  That's your way.  I don't object it.
| 1955 [austin@ha os] attribute :a, +:r, -:w
| + 1956 [transami@ru ] def +@; "#{self}_".intern; end
| + 1957 [transami@ru ] well, i just realized that the whole idea of "crazy" attributes like yours is
+ 1950 [transami@ru ] that could work, but what about non-symbol characters like white space?

YAML bug
1946 [dave@pr gp o] require 'yaml'

1958 [nobu.nokada@] Since yesterday, Proc created from the block passed with

1960 [nobu.nokada@] Current REXML::Encoding mechanism seems to have problems.

Where to install documentation
1973 [dave@pr gp o] The new RDoc/ri code needs to have a system-level directory on the
+ 1974 [jim@fr ez .o] Is an optional user directory option going to be
| 1975 [dave@pr gp o] Yup - right now though I'm asking about the standard stuff that'll get
+ 1976 [matz@ru y- a] I think data for standard tools should be somewhere under
| 1977 [dave@pr gp o] That's where I went initially, but then I got thinking. Say RubyGems
| + 1978 [jim@fr ez .o] I like the ~/.rdoc as a default for the user.
| | 1980 [austin@ha os] %USERPROFILE%/rdoc for windows users?
| | 2007 [chrismo@cl b] I'd vote against this -- everything ruby related right now has been
| + 1981 [chad@ch df w] #
| | 1983 [dave@pr gp o] The problem here is that 'ri' will have a hard job finding the most
| | 1989 [chad@ch df w] #
| + 1995 [matz@ru y- a] It should be done by RubyGems, i.e. gems installer should switch
|   2004 [gsinclair@so] That is an interesting comment that I don't fully understand.  Can you
|   + 2008 [chad@ch df w] #
|   | 2009 [matz@ru y- a] Thank you for reading my brain for me. ;-)
|   + 2010 [rich@in oe h] Well, not being Chad or Matz let me chime in since I implemented the
+ 1979 [gsinclair@so] I read $prefix as "/usr/local" or "/usr" or "C:/Ruby", etc.  That's
+ 1982 [drbrain@se m] Not on FreeBSD, documentation goes in /usr/local/share/doc/
  1984 [dave@pr gp o] Except this isn't really documentation you can read - it's a bunch of
  + 1986 [drbrain@se m] =20
  | 1990 [jlst@ge ty g] Of course, perl pod documentation is in
  + 1991 [gsinclair@so] This is like "data" for the ri "program".  An existing practice is to put
    1992 [dave@pr gp o] I have to say I really dislike this: I know it's probably "standard",
    + 1993 [gsinclair@so] [Dave:]
    | 1994 [dave@pr gp o] I think I disagree, but I'll bow to Matz.
    | 1999 [gsinclair@so] [Dave:]
    | 2001 [dave@pr gp o] The stuff in .rdoc is just RDoc documentation. The random other stuff
    | 2003 [gsinclair@so] I quite disagree.  With that reasoning, why not simply generate the RDoc
    | 2005 [gsinclair@so] As a followup point, consider this.
    + 1996 [nobu.nokada@] I feel it is good enough.
    + 1997 [chad@ch df w] #
    + 2000 [aamine@lo er] I agree with Gavin.
      2002 [dave@pr gp o] Agreed - it is currently quite messy. However, things such as
      + 2006 [ser@ge ma e-] charset="iso-8859-1"
      + 2011 [rz@li ux m6 ] one is very obvious - stuff in /usr/share can be (NFS) shared accross
      | 2012 [matz@ru y- a] Ruby scripts can be shared by NFS as well since they are non platform
      | 2017 [aamine@lo er] How do you think about security issue?
      | 2018 [matz@ru y- a] I'm not sure what you mean.  If you try to load a non program file,
      | + 2019 [nobu.nokada@] I guess they would be also better under $includedir.
      | + 2022 [aamine@lo er] Hmmm.  OK, I was too much pessimistic.  If I have freedom
      + 2020 [nobu.nokada@] $datadir is configurable regardless of $prefix and $libdir.  So
        2021 [dave@pr gp o] ...
        + 2023 [chad@ch df w] #
        | 2024 [dave@pr gp o] no - I just included the rdoc to show where the previous directory
        | 2025 [chad@ch df w] #
        + 2026 [ruby@ja es r] Are 'rdoc' directories only for the output of running rdoc?  I'm
          2027 [dave@pr gp o] Yes - they're simply for rdoc output (given that that's the tool I'm
          2028 [gsinclair@so] Has that been the premise all along?  Have you been talking only about

First pass at ri/rdoc stuff
1998 [dave@pr gp o] ...

Mixin Module, Possible Bug?
2013 [transami@ru ] "When you include a module, Ruby creates an anonymous proxy class that
2014 [matz@ru y- a] This is part of weakness of Ruby's module system.
+ 2015 [transami@ru ] Thanks matz,
| 2016 [matz@ru y- a] Inheritance/inclusion relationship is implemented by the links between
+ 2067 [transami@ru ] matz,
  2068 [matz@ru y- a] If there's any efficient way, I would.  Currently I have no idea.

rdoc fails when processing ruby stdlibs
2029 [a.bokovoy@sa] [/usr/share/ruby/1.8]$ rdoc -r
2030 [dave@pr gp o] RDoc doesn't execute the code, and so doesn't know that EXCB was

Volunteer wanted. Pay lousy
2031 [dave@pr gp o] I'm looking for a volunteer (initially just one) to add RDoc style

PATCH for Cray X1 UNICOS/mp
2032 [jcox@cr y. o] I have built Ruby on the Cray X1 supercomputer and have a patch below
+ 2033 [nobu.nokada@] It might be better to check by configure.
+ 2034 [matz@ru y- a] Ones in dln.c and numeric.c are taken.  If Nobu wants to
  2036 [jcox@cr y. o] Thanks.  Nobu's suggestion of adding to defines.h is good.  After

Syntax error in dln.c
2035 [ged@Fa ri MU] This is kind of trivial, but just wanted to let the powers that be know

--enable-pthread still segfaults...
2037 [nathaniel@ta] I've finally been able to test my application under load using the
2038 [matz@ru y- a] With --enable-pthread, longjmp() should not be called.  Show us your
2039 [nathaniel@ta] #define PACKAGE_NAME ""
2040 [matz@ru y- a] This indicates you don't have <ucontext.h> on your machine, which is
2041 [nathaniel@ta] It's in /usr/include (I'm running Debian/stable). Why would it not be
2042 [matz@ru y- a] Then show us your config.log file.
+ 2043 [nathaniel@ta] ...
| 2045 [matz@ru y- a] I'm afraid you're using old configure file.  Can you wipe off old
| 2046 [nathaniel@ta] OK, I re-copied configure from the tar archive, and it found ucontext.h
| + 2047 [matz@ru y- a] I have no idea.  Anyone?
| | 2048 [nathaniel@ta] autoconf (GNU Autoconf) 2.57
| + 2049 [nathaniel@ta] Done, and it no longer seems to segfault... now it hangs again. I've
|   2050 [matz@ru y- a] Sigh.  Show us those two files (config.h and config.log) again.
|   2051 [nathaniel@ta] Tell me about it :-(
|   2052 [matz@ru y- a] I found nothing wrong this time.  Show me the output from
|   2053 [nathaniel@ta] Well then I hope it shows the problem :-(
|   2054 [matz@ru y- a] It told me that I didn't fix your blocking problem.  I need someone to
|   2055 [nathaniel@ta] I grabbed the source package for Debian's version of libc6, and found
|   2056 [matz@ru y- a] Can you try this patch?
|   + 2059 [decoux@mo lo] Just one remark
|   | 2062 [matz@ru y- a] Does using "usleep" or "select" work on Solaris?  I.e. replace
|   | 2064 [decoux@mo lo] nasun% man usleep
|   + 2066 [nathaniel@ta] Unfortunately I won't be able to until after Christmas :-(. Did you
|     2120 [nathaniel@ta] I am extremely happy to report that the application has now run for
+ 2044 [nathaniel@ta] I tried attaching it, but just in case the attachment doesn't come