[#6548] 1.8.4 p1, warning roundup — Daniel Berger <Daniel.Berger@...>
Hi all,
[#6552] Socket Documentation — zdennis <zdennis@...>
Attached is a patch against the latest socket.c in the ruby_1_8 branch. It covers all Socket
On 11/3/05, zdennis <zdennis@mktec.com> wrote:
Gavin Sinclair wrote:
zdennis wrote:
On 11/9/05, Zach Dennis <zdennis@mktec.com> wrote:
Hi.
[#6558] Method of feeding input to regexp matching — Nikolai Weibull <mailing-lists.ruby-core@...>
I would very much like to be able to provide a Regexp object input from
[#6572] Stack trace consumes information. patch... — Hugh Sasse <hgs@...>
I have just had output like this from rails:
[#6588] Object#clone missing documentation — Eero Saynatkari <ruby-ml@...>
It appears that Object#clone, unlike Object#dup, retains
Hi,
I've attached a documentation patch which tries to address this shortcoming.
Kev Jackson wrote:
[#6602] Re: Unpack Endian Bug — "Berger, Daniel" <Daniel.Berger@...>
> -----Original Message-----
Berger, Daniel wrote:
[#6604] Sandboxing without $SAFE — why the lucky stiff <ruby-core@...>
I've been playing with Ruby sandboxing alot over the past several
[#6619] Wildness: Purpose of NOEX_PUBLIC Flag in rb_add_method? — "Charles E. Thornton" <ruby-core@...>
Several Different references to 'noex'
Charles E. Thornton wrote:
[#6625] Array::fill causes segfaults after many calls — noreply@...
Bugs item #2824, was opened at 2005-11-14 23:11
Hi,
[#6629] Strange error messages using DRb/TupleSpace — Eric Hodel <drbrain@...7.net>
Using
[#6636] alarming changes — "Ara.T.Howard" <ara.t.howard@...>
[#6639] Tuple Class — TRANS <transfire@...>
If I put together a good Tuple class for Ruby could it go into core? I
[#6650] REXML Update Please — zdennis <zdennis@...>
I submitted this as an RCR, but I didn't know that RCR's aren't for the stdlib. Matz commented on
Hi,
Yukihiro Matsumoto wrote:
[#6660] Ruby on Neko ? — Nicolas Cannasse <ncannasse@...>
Hi folks,
Nicolas Cannasse wrote:
Florian Growrote:
Nicolas Cannasse <ncannasse@motion-twin.com> writes:
On Sun, 20 Nov 2005, Christian Neukirchen wrote:
[#6672] testing for hardlink with "test(?-, ...)" flawed on Windows — noreply@...
Bugs item #2858, was opened at 2005-11-20 16:35
Hi,
--- nobuyoshi nakada <nobuyoshi.nakada@ge.com> wrote:
[#6684] semenatics of if/unless/while statement modifiers — Stefan Kaes <skaes@...>
Hi all,
On Tue, Nov 22, 2005 at 08:22:59AM +0900, Stefan Kaes wrote:
Mauricio Fern疣dez wrote:
On Nov 21, 2005, at 4:37 PM, Stefan Kaes wrote:
Eric Hodel wrote:
Hi,
Yukihiro Matsumoto wrote:
mathew wrote:
Stefan Kaes wrote:
On Tuesday 22 November 2005 12:31, Steven Jenkins wrote:
Hi --
>>>>> "m" == mathew <meta@pobox.com> writes:
Hi,
Yukihiro Matsumoto wrote:
Hi,
Yukihiro Matsumoto wrote:
Hi,
Yukihiro Matsumoto wrote:
On Nov 21, 2005, at 9:37 PM, Stefan Kaes wrote:
Eric Hodel wrote:
URABE Shyouhei wrote:
On Tue, 22 Nov 2005, Stefan Kaes wrote:
Ara.T.Howard wrote:
Hi --
David A. Black wrote:
Hi --
David A. Black wrote:
Hi --
David A. Black wrote:
Hi -
On Tuesday 22 November 2005 15:37, David A. Black wrote:
Hi --
On Tue, 22 Nov 2005, Stefan Kaes wrote:
Mathieu Bouchard wrote:
[#6721] String#index does not work correctly on SuSE10.0 x86_64 — "Kanis, Lars" <Kanis@...>
Hi folks,
[#6798] ruby 1.8.4 preview2 — Yukihiro Matsumoto <matz@...>
Hi,
On Nov 30, 2005, at 8:03 AM, Yukihiro Matsumoto wrote:
>>>>> "E" == Eric Hodel <drbrain@segment7.net> writes:
On Dec 4, 2005, at 4:07 AM, ts wrote:
>>>>> "E" == Eric Hodel <drbrain@segment7.net> writes:
On 11/30/05, Yukihiro Matsumoto <matz@ruby-lang.org> wrote:
Hi,
Yukihiro Matsumoto wrote:
Hi.
Re: [ ruby-Bugs-2715 ] [PATCH] 1.8.3 ruby.c doesn't compile on OS X due to missing char **environ
Yukihiro Matsumoto wrote:
> Hi,
>
> In message "Re: [ ruby-Bugs-2715 ] [PATCH] 1.8.3 ruby.c doesn't compile on OS X due to missing char **environ"
> on Tue, 25 Oct 2005 15:09:54 +0900, noreply@rubyforge.org writes:
>
> |I'm working on getting Ruby 1.8.3 into Fink, the Debian style
> |package managed system for Mac OS X.
> |
> |Ruby 1.8.3 introduced the usage of the
> |
> |extern char **environ;
> |
> |in ruby.c.
> |
> |Mac OS X does not have this symbol, which you can confirm by greping
> |through /usr/include and nm on the /usr/lib/*.
> |
> |The work around is to do something like this:
>
> I'm not sure if it's OK to replace environ by _NSGetEnviron() since
> this code assumes UNIX style environment variable memory map. Could
> anyone confirm this is OK or not for Mac OS X?
>
> matz.
Hello,
This is a follow up note to all the replies to my original bug report.
Let me provide a little more background information. I'm one of the Fink
maintainers that's supporting the Ruby package and upgrading it from 1.8.1 to 1.8.3.
One of the differences between the vanilla build and the Fink build is that the
vanilla build uses this link:
$ cc -dynamiclib -undefined suppress -flat_namespace -install_name
/tmp/blair/rrr-1/lib/libruby.dylib -current_version 1.8.3 -compatibility_version
1.8 ........ -o libruby.1.8.3.dylib
while Fink does not suppress undefined symbols and uses this:
$ cc -dynamiclib -install_name /sw/lib/libruby.1.8.dylib -current_version 1.8.3
-compatibility_version 1.8 ........ -o libruby.1.8.3.dylib
With the Fink link, I get
ld: Undefined symbols:
_environ
/usr/bin/libtool: internal link edit command failed
make: *** [libruby.1.8.3.dylib] Error 1
Hence my original patch to change the environ.
Regarding _NSGetEnviron and the memory map, I don't know, but you're already
using this in several places in ruby:
eval.c:
#if defined(__APPLE__)
#define environ (*_NSGetEnviron())
#elif !defined(_WIN32) && !defined(__MACOS__) || defined(_WIN32_WCE)
extern char **environ;
#endif
char **rb_origenviron;
hash.c:
#ifdef _WIN32
#define GET_ENVIRON(e) (e = rb_w32_get_environ())
#define FREE_ENVIRON(e) rb_w32_free_environ(e)
static char **my_environ;
#undef environ
#define environ my_environ
#elif defined(__APPLE__)
#undef environ
#define environ (*_NSGetEnviron())
#define GET_ENVIRON(e) (e)
#define FREE_ENVIRON(e)
#else
extern char **environ;
#define GET_ENVIRON(e) (e)
#define FREE_ENVIRON(e)
#endif
If I may suggest, maybe this common code for environ could be moved into a .h file.
Another point, the Fink package patches the builds to remove the '-undefined
suppress' from all links, including building the extension bundles. Would it be
a good idea to remove this linker option, to catch bugs like this?
<speculation>
I'm not exactly clear why this works in the vanilla build even when ruby.c
declares char **environ as extern, but the symbol does appear to be created, as
it exists in the resulting ruby binary that is dynamically linked against
libruby.dylib:
$ nm /tmp/blair/rrr-1/bin/ruby |grep environ
00002008 D _environ
$ otool -L /tmp/blair/rrr-1/bin/ruby
/tmp/blair/rrr-1/bin/ruby:
/tmp/blair/rrr-1/lib/libruby.dylib (compatibility version 1.8.0,
current version 1.8.3)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
version 71.1.4)
/usr/lib/libobjc.A.dylib (compatibility version 1.0.0, current version
227.0.0)
</speculation>
Regards,
Blair
--
Blair Zajac, Ph.D.
<blair@orcaware.com>
Subversion and Orca training and consulting
http://www.orcaware.com/svn/