22013-22844

21819-23811 subjects 22249-23532

[Bug #1144] Tag filename problem in the SVN Repo on Windows (git import)
22013 [redmine@ru y] Bug #1144: Tag filename problem in the SVN Repo on Windows (git import)

[Bug #1145] IRB adds trailing quote to wtring methods when performing auto-completion
22018 [redmine@ru y] Bug #1145: IRB adds trailing quote to wtring methods when performing auto-completion

[Feature #1147] File.expand_path('~') should honor ENV['USERPROFILE'] on Windows
22026 [redmine@ru y] Feature #1147: File.expand_path('~') should honor ENV['USERPROFILE'] on Windows

[Feature #1147] File.expand_path('~') should honor ENV['USERPROFILE'] on Windows
22027 [redmine@ru y] Issue #1147 has been updated by Suraj Kurapati.

[Bug #1116](Assigned) Compilation error for Solaris 10 sparc
22029 [redmine@ru y] Issue #1116 has been updated by Nobuyoshi Nakada.

[Bug #1120](Closed) [PATCH] fix build on GNU/kOpenSolaris
22030 [redmine@ru y] Issue #1120 has been updated by Nobuyoshi Nakada.

[Bug #1149] gem update --system failed
22032 [redmine@ru y] Bug #1149: gem update --system failed

1.9.1 & Kernel.load(..., true) --> scope problem?
22033 [johan556@gm ] I recently tried to use MRI 1.9.1 for one of my existing scripts.

[Bug #1149](Rejected) gem update --system failed
22036 [redmine@ru y] Issue #1149 has been updated by Nobuyoshi Nakada.

[Bug #835] fix for Documentation on ruby*/lib/benchmark
22037 [redmine@ru y] Issue #835 has been updated by Akinori MUSHA.

[Bug #1151] Aliased methods change super logic when retrieved with Object#method
22040 [redmine@ru y] Bug #1151: Aliased methods change super logic when retrieved with Object#method

Behavior of super against undef'ed methods?
22045 [charles.nutt] There some peculiarity with how undef, class_eval, and super are

[Bug #1152] profiler.rb is not concurrent-execution threadsafe
22046 [redmine@ru y] Bug #1152: profiler.rb is not concurrent-execution threadsafe

Seeing threading issues on PowerPC OS X
22047 [dave@pr gp o] When I built the PickAxe on my Intel OS X box, I used to see a number

1.8.6 FTP (was Re:  Re: 1.8.7 Specifics)
22050 [dooby@d1 .k ] As an official Ruby, could ruby-lang ftp be kept recent?

A short memorandum (Re: 1.8.7 Specifics)
22051 [shyouhei@ru ] Let me leave a memo to remember issues I can think of.
+ 22054 [charles.nutt] I strongly agree here. Github would be great if this were intended to be
| 22055 [charles.nutt] A shorter way to say what I mean: 1.8.6 should not move without all of
| 22079 [ruby-ml@ki t] I will agree and disagree a bit: keeping the distribution at
| + 22091 [charles.nutt] In any case, since there's no realistic way to mirror a github repo from
| | + 22093 [ruby-ml@ki t] What makes you say that? For reference, here is the git-svn
| | | + 22099 [charles.nutt] I don't know how to make git-svn pushable to a remote repo and pushable
| | | | 22104 [gwtmp01@ma .] I'm not sure I see the problem. The only people that need bidirectional
| | | | 22109 [charles.nutt] Now to clarify, I don't see any issue if svn and github are still both
| | | + 22103 [ruby-ml@ki t] 1. Import code from SVN trunk to the Git repo.
| | + 22101 [shyouhei@ru ] Maybe stewardship and source code hosting service are different stories.  EY
| |   22107 [ezmobius@gm ] Shyouhei-
| |   22123 [shyouhei@ru ] Well I'm maintaining a list of revisions that needs/experienced backport to
| |   22128 [nobu@ru y- a] And I'm still awaiting the newer patches for 1.8 head, but now
| |   22132 [brent@mb ri ] Nobu and Urabe,
| |   + 22133 [vjoel@pa h. ] A minor comment: #increase is a verb as well as a noun, and my first
| |   | 22135 [brent@mb ri ] You have a point there.
| |   + 22139 [charles.nutt] These expose implementation-specific details of GC. Should they perhaps
| |     22145 [brent@mb ri ] Charlie,
| |     22152 [charles.nutt] I agree that the GC methods are all generally implementation-specific,
| |     + 22155 [vjoel@pa h. ] What about putting these methods in MRI::GC or GC::MRI? That would fail
| |     + 22156 [brent@mb ri ] Charlie,
| |     | 22164 [charles.nutt] The problem is an ever-growing set of methods you have to check for. If
| |     | + 22166 [hramrach@ce ] And if you are dealing with different versions of the same
| |     | | 22167 [charles.nutt] Splitting them into at least one extension would at least isolate it.
| |     | | 22169 [hramrach@ce ] What I have seen proposed is splitting into one extension per
| |     | + 22168 [brent@mb ri ] Charlie,
| |     + 22157 [hongli@pl n9] I understand that this is frustrating to you as a Ruby implementer, but
| |       22162 [charles.nutt] I never said don't expose them...just don't make them part of the
| |       22171 [ruby-ml@ki t] I would prefer having to `require "gc"` for the insulation. Only
| |       22172 [charles.nutt] I'd find that totally satisfactory too. The implication there would be
| + 22094 [brent@mb ri ] ====
|   22097 [charles.nutt] git-svn works great on a local repository (I use it myself), but I know
|   + 22105 [ruby-ml@ki t] I think your expectations are set based on a centralized VCS?
|   + 22106 [brent@mb ri ] Charlie,
|     22110 [charles.nutt] The problem is that 1.8.6 is not supposed to accept commits that haven't
|     22114 [brent@mb ri ] Charlie,
|     22120 [charles.nutt] 1_8_6 is already its own branch and only gets merged as the maintainer
+ 22056 [wycats@gm il] Out of curiosity, which backportings are these?

[Bug #1140] Parsing error gives method_missing a nonexistant method name
22052 [redmine@ru y] Issue #1140 has been updated by Ben Rayfield.

[Bug #738] Repeated calls to popen cause thread problems
22053 [redmine@ru y] Issue #738 has been updated by Michal Suchanek.

[Bug #1157] missing zlib.rb?
22058 [redmine@ru y] Bug #1157: missing zlib.rb?
22087 [rogerdpack@g] yeah 1.9 should probably complain more when there's no zlib or no

[Bug #1157] missing zlib.rb?
22059 [redmine@ru y] Issue #1157 has been updated by Fred Obermann.
22063 [nobu@ru y- a] You need the package for development too.

[Bug #1158] make appears to hang on compilation of  parse.c  with gcc version 4.3.2
22060 [redmine@ru y] Bug #1158: make appears to hang on compilation of  parse.c  with gcc version 4.3.2
22064 [nobu@ru y- a] It doesn't "hang" but just takes several minutes, and gcc's

[Bug #1157] missing zlib.rb?
22061 [redmine@ru y] Issue #1157 has been updated by Fred Obermann.
22072 [neleai@se na] Try build ext/zlib. And its zlib.so no zlib.rb which is required.

Dir.glob and duplicates?
22065 [charles.nutt] I was fixing a JRuby Dir.glob spec failure where we produced a duplicate
+ 22066 [akr@fs j. rg] It depends on the pattern.  [ruby-core:16853]
| 22067 [charles.nutt] So if some globs may produce dupes, then it seems that JRuby producing
| 22068 [akr@fs j. rg] Would you show us the pattern?
| 22069 [charles.nutt] In the JRuby dupe case it's 'a/**/b/**/e'.
| 22070 [akr@fs j. rg] It has no braces.  So the result should not have dupes.
| + 22083 [matz@ru y- a] Because glob returns existing file names only unlike shells.
| | 22095 [dooby@d1 .k ] So, if the *nix shells are not emulated exactly, we should
| + 22089 [charles.nutt] I would argue that since glob can be implemented in different ways, that
|   22098 [akr@fs j. rg] The simple "may return dupes" specification breaks
|   22111 [charles.nutt] Well, if there are known issues with glob returning dupes, perhaps the
|   22112 [akr@fs j. rg] I don't think it is a bug.
|   22121 [dooby@d1 .k ] Maybe symlinks can produce real dup's ?
|   22138 [charles.nutt] I guess my question stands...if I fixed JRuby to never return
|   22144 [dooby@d1 .k ] Sadly, from what I've read, I think you would be breaking specified MRI
+ 22071 [nobu@ru y- a] The pattern at line 204 is '**/*fil?{,.}*', which is equivalent
  + 22074 [dooby@d1 .k ] Ha! A puzzle :)
  | 22081 [pit.capitain] Thanks daz, I didn't get it at first. Now I understand the example.
  | 22084 [ruby-ml@ki t] Yeah, definitely not the clearest of usages, but that is
  + 22090 [charles.nutt] I think the presence of duplicates is a glob implementation detail that
    + 22108 [brixen@gm il] On Feb 14, 10:47=A0am, Charles Oliver Nutter <charles.nut...@sun.com>
    + 22175 [ryand-ruby@z] I disagree. I think the behavior as specced is not implementation
      22176 [charles.nutt] And why should dupes be specified behavior? Why would no dupes be a
      + 22178 [ryand-ruby@z] I haven't the faintest why you'd ask these questions of me. They're
      | 22188 [charles.nutt] I asked you because you responded with your opinion. So I asked you for
      + 22185 [Thomas.Enebo] My stab at this is because most existing glob implementations do this
        + 22186 [ jfh@uf .e u] Interesting -- being a zsh user I've always thought of braces differently
        + 22189 [charles.nutt] Yeah, the least surprising result is rather surprising, since I haven't

Re: MBARI patches on GitHub
22073 [brent@mb ri ] Michael,
+ 22102 [stephen.bann] Brent, thanks for putting them on github. This makes it very easy to
| 22129 [brent@mb ri ] Stephan,
+ 22113 [rogerdpack@g] I was able to fork and build it quite easily, too.  I have a few
  22119 [brent@mb ri ] Roger,

[Bug #1001](Closed) Interpreter-wide deadlock when native close() is called on fd which is currently being IO#select()ed
22076 [redmine@ru y] Issue #1001 has been updated by Akira Tanaka.

[Bug #1053](Closed) Obsolete methods in Pathname
22077 [redmine@ru y] Issue #1053 has been updated by Akira Tanaka.

[Bug #883] Failure: test_handle_special_CROSSREF_no_underscore(TestRDocMarkupToHtmlCrossref)
22078 [redmine@ru y] Issue #883 has been updated by Tomoyuki Chikanaga.

[Feature #1141](Rejected) assignment of variable in "right" if statement fails
22080 [redmine@ru y] Issue #1141 has been updated by Yukihiro Matsumoto.

File#size et al
22088 [rogerdpack@g] In a recent discussion[1] on some more "convenience" methods for File,
22196 [matz@ru y- a] Currently we have File#stat and #atime, #mtime, #ctime.  What else do
22313 [rogerdpack@g] Well I'm not saying the current setup doesn't work, or trying to be

[Bug #1160] tracer attempts to show lines from binary files
22096 [redmine@ru y] Bug #1160: tracer attempts to show lines from binary files
22844 [rogerdpack@g] Any ideas here?

[Bug #1161] Build Assertion Failure with VC+++ - Unitialized variable in rb_define_hooked_variable
22115 [redmine@ru y] Bug #1161: Build Assertion Failure with VC+++ - Unitialized variable in rb_define_hooked_variable

[Bug #1162] Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22116 [redmine@ru y] Bug #1162: Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22127 [nobu@ru y- a] It's false assertion.  Report it to Microsoft.

[Bug #1161] Build Assertion Failure with VC+++ - Unitialized variable in rb_define_hooked_variable
22117 [redmine@ru y] Issue #1161 has been updated by Charlie Savage.

[Bug #1162] Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22118 [redmine@ru y] Issue #1162 has been updated by Charlie Savage.

[Feature #934](Closed) request for Time.parse to no longer return Time.now on failure
22122 [redmine@ru y] Issue #934 has been updated by Akira Tanaka.

Aliased names show in backtrace in 1.8, but not 1.9
22124 [charles.nutt] Is this an intentional change?

[Bug #1161](Closed) Build Assertion Failure with VC+++ - Unitialized variable in rb_define_hooked_variable
22126 [redmine@ru y] Issue #1161 has been updated by Nobuyoshi Nakada.

A message from the maintainer of the 1.8.7+ series
22130 [akinori.mush] As I don't feel like sounding very constructive or informative by
+ 22131 [zznmeb@gm il] Thanks!! I'm solidly in the "make all my code work on 1.9.1 and never
+ 22149 [dblack@ru yp] Thanks for the clear statement of where we stand. The main thing is

[Bug #1162] Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22134 [redmine@ru y] Issue #1162 has been updated by Charlie Savage.
22150 [nobu@ru y- a] Yes, it's a reasonable and expected behavior.

Confused about some code in mathn.rb
22136 [charles.nutt] loop do
22137 [charles.nutt] More confusion in the same method...if I comment out the loop, it
22140 [charles.nutt] FWIW, this code seems to be the same on 1_8 branch. If I comment out the
22141 [charles.nutt] ruby_bug "#175", "1.8.7" do
22153 [charles.nutt] Ok, so it seems like the "neaed" typo was corrected, and then the bug

[Patch 191p0 ] for ostruct freeze behavior
22142 [robert.dober] Please find attached a patch for ostruct to behave frozen for already
+ 22143 [robert.dober] Sorry for replying on my own post.
+ 22154 [vjoel@pa h. ] Can you delegate #freeze to the @table, or is it more complicated?
  22160 [robert.dober] I respected the original design, and that was of course simpler, a one
  22161 [robert.dober] No I am not, you did, sorry I had the impression that was my patch, stupid me.
  22300 [robert.dober] Do I have to file a bug report?
  22307 [robert.dober] When I read my post I was shocked, but this was a sincere question. I
  22419 [robert.dober] I was about to file an error report.

[Bug #1162](Closed) Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22151 [redmine@ru y] Issue #1162 has been updated by Nobuyoshi Nakada.

[Bug #657] Thread.new { fork }
22158 [redmine@ru y] Issue #657 has been updated by James M. Lawrence.
22183 [matz@ru y- a] It doesn't happen on my

Thread#abort_on_exception behavior
22159 [quixoticsyco] Is this difference between 1.8 and 1.9 expected?
22163 [quixoticsyco] I intended to revert this unwind-via-throw experiment before sending.

[Bug #1162] Build Assertion Failure with VC+++ - Incorrect flushing of stdout/stderr
22170 [redmine@ru y] Issue #1162 has been updated by Charlie Savage.
22343 [nobu@ru y- a] It is a compile time flag and irrelevant to debug built msvcrt.

Modification during Iteration
22173 [gavin@ph og ] It has recently come to my attention that the following code causes an

[Feature #1144](Closed) Tag filename problem in the SVN Repo on Windows (git import)
22179 [redmine@ru y] Issue #1144 has been updated by Nobuyoshi Nakada.

[Bug #1164] 1.9/windows memroy leak with rand() AND inspect
22184 [redmine@ru y] Bug #1164: 1.9/windows memroy leak with rand() AND inspect
+ 22192 [nobu@ru y- a] I tried it with 1.9.1dev but memory usage displayed by
+ 22220 [rogerdpack@g] I was able to recreate it with

[Bug #657] Thread.new { fork }
22187 [redmine@ru y] Issue #657 has been updated by James M. Lawrence.
22191 [nobu@ru y- a] I could reproduce it, forked process never die.  I'll
22204 [nobu@ru y- a] Sorry, forgotten to commit an old fix.

[Bug #1165] Range.eql? and Range.== bug with subclasses
22190 [redmine@ru y] Bug #1165: Range.eql? and Range.== bug with subclasses

Pragmas?
22193 [zznmeb@gm il] constant. The program compared some data with this constant and took a
22194 [nobu@ru y- a] 1.8.7 supports all of 1.8.6 syntax and almost of the semantics
22195 [gregory.t.br] I think this could be very useful.  Actually, if we had something like

[Bug #1166] Simple gcc-4.4 fix
22197 [redmine@ru y] Bug #1166: Simple gcc-4.4 fix

[Bug #1166](Closed) Simple gcc-4.4 fix
22198 [redmine@ru y] Issue #1166 has been updated by Nobuyoshi Nakada.

[Bug #1091] possible bad handling of return value of OCSP_basic_verify in ext/openssl/ossl_ocsp.c
22199 [redmine@ru y] Issue #1091 has been updated by Laurent Sansonetti.
22205 [nobu@ru y- a] Do you mean that OCSP_basic_verify() returns positive value on

MBARI patches on 1.8.6p287
22200 [brent@mb ri ] in the branch v1_8_6_287-mbari
+ 22217 [kingmt@gm il] I will try and take a look at this soon, hopefully before the end of the
+ 22221 [rogerdpack@g] does the -fno-stack-protector stuff make much of a speed difference?
  22231 [brent@mb ri ] I used to think it was more, but in fact, -fno-stack-protector probably saves

[Bug #1167] Regexp: Stack overflow in regexp matcher
22201 [redmine@ru y] Bug #1167: Regexp: Stack overflow in regexp matcher

ruby-1.9.1-p0 build failure on i586
22202 [kanarip@fe o] I'm the ruby maintainer for the Fedora Project, and as such I was going

[Bug #657](Closed) Thread.new { fork }
22203 [redmine@ru y] Issue #657 has been updated by Nobuyoshi Nakada.

ruby-1.9.1-p0 build failure on i586
22206 [kanarip@fe o] I'm the ruby maintainer for the Fedora Project, and as such I was going
+ 22224 [rogerdpack@g] What machine is this on exactly? And it reproduces easily?
| 22263 [kanarip@fe o] This is on one of many buildhosts, but it occurs on my local machine as
| 22272 [ko1@at ot ne] How to (try|debug) it?
| 22274 [kanarip@fe o] ...
| 22275 [ko1@at ot ne] Should I prepare newest Fedora machine to debug it?  I only have a
| 22278 [kanarip@fe o] There's no public machines available where you can just log in and do stuff.
+ 22247 [nobu@ru y- a] What version of gcc?
| 22264 [kanarip@fe o] gcc-4.4.0, according to
+ 22585 [akr@fs j. rg] I installed Fedora 11 Alpha.
  22671 [kanarip@fe o] Have you installed the -debuginfo packages?
  22672 [akr@fs j. rg] Yes.

[Bug #1157] missing zlib.rb?
22207 [redmine@ru y] Issue #1157 has been updated by Michael Lang.

[Bug #1169] Thread#priority still doesn't work
22208 [redmine@ru y] Bug #1169: Thread#priority still doesn't work
22235 [brent@mb ri ] Native threaded implementations will have a difficult time implementing the

[Bug #1170] test_file_exhaustive.rb fails if building as root
22209 [redmine@ru y] Bug #1170: test_file_exhaustive.rb fails if building as root
+ 22222 [rogerdpack@g] why do they fail?
+ 22227 [mame@ts .n .] Done.  Thank you!

[Bug #1091] possible bad handling of return value of OCSP_basic_verify in ext/openssl/ossl_ocsp.c
22210 [redmine@ru y] Issue #1091 has been updated by Laurent Sansonetti.

[Bug #1171] ri: if "pager" exists, use it before trying less.
22211 [redmine@ru y] Bug #1171: ri: if "pager" exists, use it before trying less.

[Bug #1172] [sparc] *** glibc detected *** ruby1.9: free(): invalid pointer: 0xf7ef6a54 ***
22212 [redmine@ru y] Bug #1172: [sparc] *** glibc detected *** ruby1.9: free(): invalid pointer: 0xf7ef6a54 ***

[Bug #1173] Method name typo in soap/mimemessage.rb
22214 [redmine@ru y] Bug #1173: Method name typo in soap/mimemessage.rb

[Feature #934] request for Time.parse to no longer return Time.now on failure
22216 [redmine@ru y] Issue #934 has been updated by Roger Pack.

[Bug #1170](Closed) test_file_exhaustive.rb fails if building as root
22228 [redmine@ru y] Issue #1170 has been updated by Yusuke Endoh.

[Bug #1174] compilation error with OpenSSL 0.9.8j
22229 [redmine@ru y] Bug #1174: compilation error with OpenSSL 0.9.8j

[Bug #1164] 1.9/windows memroy leak with rand() AND inspect
22241 [redmine@ru y] Issue #1164 has been updated by Usaku NAKAMURA.
22302 [rogerdpack@g] As a note, compiled with Visual Studio 2008 shows it, as well.

[Bug #1173] Method name typo in soap/mimemessage.rb
22242 [redmine@ru y] Issue #1173 has been updated by Akinori MUSHA.

[Bug #1173](Closed) Method name typo in soap/mimemessage.rb
22243 [redmine@ru y] Issue #1173 has been updated by Akinori MUSHA.

[Bug #1173](Open) Method name typo in soap/mimemessage.rb
22244 [redmine@ru y] Issue #1173 has been updated by Akinori MUSHA.

[Bug #1176] uninitialized constant Zlib::FFI after fresh ruby compilation
22245 [redmine@ru y] Bug #1176: uninitialized constant Zlib::FFI after fresh ruby compilation

YASNP (Yet Another Selector Namespace Proposal)
22246 [wycats@gm il] The idea is to make selectors like optional versions of Python imports.
+ 22248 [micathom@gm ] Just out of curiosity: Wouldn't the problem with your examples be that
| 22250 [wycats@gm il] Actually it's completely unrelated.
+ 22251 [charles.nutt] The idea is that at parse time it saves off the node associated with
| 22252 [charles.nutt] * This version searches the namespace for the method *unconditionally*.
| 22262 [flo@an er gr] That sounds a lot like the Objective-C method poseAs, discarding some
| + 22266 [charles.nutt] I see the namespacing more like adding an extra layer of indirection to
| + 22267 [wycats@gm il] 2009/2/19 Florian Gilcher <flo@andersground.net>
|   + 22270 [charles.nutt] I suppose there's another argument in favor of an explicit bounding of
|   | 22273 [wycats@gm il] I'd be perfectly happy to have it take a block.
|   + 22285 [flo@an er gr] ...
|     22295 [wycats@gm il] Ok, based on a bunch of comments I got from Aaron Patterson and John
|     + 22296 [charles.nutt] I'm trying to work out the details in my head. The main complication I
|     + 22316 [jim.weirich@] +1
|     | 22322 [aaron@te de ] +1 as well, but I like the block.  For me, it seems to be more clear as
|     | + 22323 [wycats@gm il] The thing is that it's not a *real* block, but that might not matter
|     | + 22330 [ruby-ml@ki t] I am not sure I like this plan at all, really: it would be better
|     | | 22369 [Daniel.Berge] =20
|     | | 22372 [wycats@gm il] Can you explain *why* you don't like it?
|     | | + 22374 [flo@an er gr] ...
|     | | | 22375 [wycats@gm il] The people who would use it (i.e. frameworks and libraries) would probably
|     | | + 22377 [ruby-ml@ki t] It is difficult to see the extent of changes. I prefer the method-wise
|     | | | 22379 [wycats@gm il] I'd be fine with "foo".Merb:to_url as one way of accessing the namespace,
|     | | | 22394 [djberg96@gm ] I'm not saying it might not be useful, I just want a more fine grained
|     | | | 22396 [wycats@gm il] As long as that syntax is technically parsable, it's fine with me.
|     | | + 22378 [Thomas.Enebo] Famous last words...One poignant blog entry can undo your statement in a
|     | |   22380 [wycats@gm il] 2009/2/23 Thomas Enebo <Thomas.Enebo@sun.com>
|     | + 22409 [matz@ru y- a] +1 to have some kind of selector namespace (it's planned for 2.0
|     |   + 22422 [wycats@gm il] Matz,
|     |   + 22423 [charles.nutt] I think it's impossible to have per-file without making "use" a special
|     |   | + 22424 [wycats@gm il] I don't think it's important that it be at the top of the file, but it does
|     |   | | 22426 [neleai@se na] +1 to per-file
|     |   | | 22430 [charles.nutt] This could certainly be simplified. I'm also leaning toward "using" and
|     |   | + 22476 [brent@mb ri ] There may be a middle way
|     |   |   + 22484 [ruby-ml@ki t] ...Why oh why? :)
|     |   |   | 22493 [charles.nutt] Yes, +1 for using ... end. If we go with explicit scoping (or maybe make
|     |   |   + 22487 [ola.bini@gm ] Of course, this change would mask many legitimate errors where you
|     |   + 22427 [brixen@gm il] ce =A0 =A0 =A0 =A0Proposal)"
|     |     + 22433 [ruby-ml@ki t] In addition to overriding, selectors can be used for adding (and,
|     |     | + 22435 [wycats@gm il] 2009/2/24 Eero Saynatkari <ruby-ml@kittensoft.org>
|     |     | | 22441 [brixen@gm il] are
|     |     | | 22442 [wycats@gm il] 2009/2/24 Brian Ford <brixen@gmail.com>
|     |     | | 22446 [jdeville@mi ] I agree that this will be used in ways other than just framework creators. =
|     |     | | + 22447 [charles.nutt] Going out and changing (monkeypatching) libraries (or namespaces)
|     |     | | + 22448 [jim.weirich@] I must admit, I'm a little confused by the above.  Maybe we have
|     |     | |   22449 [jdeville@mi ] Clarification of my example: If three namespaces define xyz, and you want a=
|     |     | |   + 22450 [wycats@gm il] I'm also in favor of discussing this, but all I hear so far in opposition is
|     |     | |   | + 22460 [flo@an er gr] ...
|     |     | |   | | + 22466 [hramrach@ce ] This is a clear use case for fine-grained control of namespaces. You
|     |     | |   | | | 22486 [charles.nutt] I'm starting to think that if we don't have explicitly bounded
|     |     | |   | | + 22469 [wycats@gm il] 2009/2/25 Florian Gilcher <flo@andersground.net>
|     |     | |   | | | 22485 [flo@an er gr] ...
|     |     | |   | | | 22491 [wycats@gm il] 2009/2/25 Florian Gilcher <flo@andersground.net>
|     |     | |   | | + 22470 [matz@ru y- a] Good point.
|     |     | |   | | + 22483 [charles.nutt] I think the FUD here is "oh no, what if this happens or that happens and
|     |     | |   | |   + 22492 [flo@an er gr] That may be true or not, but it is still against the rules of a good
|     |     | |   | |   | 22496 [charles.nutt] Yeah, I don't see it as being too difficult, though IRB does force the
|     |     | |   | |   + 22530 [duerst@it ao] This might be a very silly question, but how does the module/namespace
|     |     | |   | |     22531 [wycats@gm il] module Foo; def to_s; 'blah'; end; end
|     |     | |   | + 22471 [brixen@gm il] is
|     |     | |   |   + 22474 [radek.bulat@] Yehuda, I wonder on one thing. If you want in your framework/library
|     |     | |   |   | 22478 [wycats@gm il] (str)))
|     |     | |   |   | + 22481 [flo@an er gr] Yehuda is correct there. Thats the stuff you need an IDE and imports
|     |     | |   |   | + 22499 [jim.weirich@] Plus this technique wouldn't allow polymorphic methods based on the
|     |     | |   |   |   22500 [charles.nutt] GREAT point...I hadn't even considered that. That's a much stronger
|     |     | |   |   + 22480 [wycats@gm il] 2009/2/25 Brian Ford <brixen@gmail.com>
|     |     | |   |   + 22488 [martindemell] I'm not seeing your point - selector namespaces aren't a fence to keep
|     |     | |   |   + 22505 [jim.weirich@] You should use a namespace whenever you add a method to a module or
|     |     | |   |   | 22565 [charles.nutt] I thought I'd sent this before...hopefully it's still relevant.
|     |     | |   |   | + 22571 [ruby-ml@ki t] Not sure this is the best use case?
|     |     | |   |   | | 22575 [charles.nutt] Since it needs to define a new constant, it can't be called against a
|     |     | |   |   | + 22572 [gwtmp01@ma .] I think it is helpful to have concrete examples like this.
|     |     | |   |   | + 22616 [duerst@it ao] I'm not exactly sure I'd prefer
|     |     | |   |   + 22548 [brixen@gm il] n is
|     |     | |   |     + 22552 [jdeville@mi ] This is actually another important case to consider. We've stated that name=
|     |     | |   |     | 22553 [wycats@gm il] Agreed that it should be LIFO, in order to be consistent with existing Ruby
|     |     | |   |     + 22570 [hramrach@ce ] I assumed you could just type
|     |     | |   |       22574 [wycats@gm il] using Merb::*
|     |     | |   + 22451 [jim.weirich@] That's why Charlie's prototyping is so important.  Once we have real
|     |     | |     22452 [charles.nutt] If folks are interested in playing with it more, I could expand it to
|     |     | |     22453 [wycats@gm il] I'd like that a lot :)
|     |     | |     22455 [charles.nutt] ...
|     |     | + 22436 [charles.nutt] Unrequiring and unextending make global changes. I think the selector
|     |     + 22457 [matz@ru y- a] Interesting.  I thought people behind Rubinius would love (or even
|     |       + 22459 [wycats@gm il] To be honest, I'm not sure what Rubinius' experience has anything to do with
|     |       | 22489 [charles.nutt] I think Matz is referring to a desire to ensure that within Rubinius's
|     |       + 22462 [ruby-ml@ki t] Selector namespaces was and probably still is a planned feature; if
|     |       + 22473 [brixen@gm il] ce Proposal)"
|     |         22494 [charles.nutt] So on the one hand you support Rubinius being totally open, even if that
|     + 22438 [charles.nutt] ...
|       + 22440 [wycats@gm il] Big +1 Charlie!
|       + 22454 [neleai@se na] This code should output inside namespace: inside namespace: inside namespace: inside namespace: inside namespace: inside namespace: ...
|         22456 [charles.nutt] The namespace applies from that point forward; the original Foo::Bar
+ 22490 [ola.bini@gm ] I have read the full thread now, and have some concrete questions about
| + 22495 [matz@ru y- a] Here's MY opinion.
| | + 22498 [david@da id ] I have not been following this thread, and have no detailed
| | + 22506 [ola.bini@gm ] OK, my main question about how this should work then is basically that
| | | 22510 [charles.nutt] They certainly could be names, or the namespaces could be declared in a
| | | 22514 [ola.bini@gm ] Well, my point was that in that case namespaces/selectors isn't strictly
| | | + 22519 [Thomas.Enebo] Heavy?  or just an additional cost?  All nodes in that file will need to
| | | + 22521 [charles.nutt] Yes, sounds like you're thinking along the same lines I am..."SEL_CALL"
| | |   22522 [gwtmp01@ma .] During this modified method lookup process...
| | |   22523 [wycats@gm il] 2009/2/25 Gary Wright <gwtmp01@mac.com>
| | |   + 22524 [charles.nutt] In my prototype, they would win as well, but only because it expects the
| | |   | 22525 [gwtmp01@ma .] It seems to me that 'super' behaves as a resumption of the method
| | |   + 22526 [jim.weirich@] My take on this is that selector namespaces are for preventing
| | |     22527 [gwtmp01@ma .] In Ruby's normal method invocation process, the runtime searches a
| | |     22528 [jim.weirich@] Which namespaces?  The ones in scope at the original call site, or the
| | |     22529 [gwtmp01@ma .] My first thought was the one in scope at the original call site.  I'll
| | |     + 22532 [charles.nutt] There are several contortions involved in super calls, definitely.
| | |     + 22533 [jim.weirich@] Here's a concrete example of why super searching namespaces at the
| | |       22537 [charles.nutt] I've talked about this a bit with Yehuda offline and super is definitely
| | |       + 22542 [gwtmp01@ma .] I'm still trying to work out the ramifications of Yehuda's proposal.
| | |       | 22547 [hramrach@ce ] For me it looks like selector namespaces get us some of the problems
| | |       + 22544 [ruby-ml@ki t] This is one of the reasons why I think unrequire/uninclude/unextend
| | |         22554 [hramrach@ce ] I agree except for the "in place of separate selector namespaces".
| | + 22518 [charles.nutt] Very difficult. Perhaps we're thinking about namespaces differently? For
| | + 22540 [halostatue@g] Is there any reason that the above can't be made a valid syntax? I
| |   22549 [charles.nutt] obj.< is a valid call
| + 22501 [jim.weirich@] (1) If two selector namespaces are in scope, and they both define a
| | + 22502 [charles.nutt] In my prototype they're first-come, first-served. You can have many
| | | + 22507 [wycats@gm il] I think it's likely the case that explicit calls to namespaced methods
| | | | 22511 [charles.nutt] Yeah, that shouldn't be hard; send_with_namespace would just explicitly
| | | + 22541 [halostatue@g] od
| | |   22550 [charles.nutt] My brain works FIFO I guess. I read this as "check Foo first, then check
| | |   + 22555 [gwtmp01@ma .] I find the FIFO ordering strange since it seems to go against the
| | |   | 22557 [charles.nutt] Without an actual scope, I think it could be confusing either way. I'm
| | |   + 22563 [halostatue@g] I don't think that I could come up with a convincing proposal because
| | + 22503 [ola.bini@gm ] Good questions. At this point I'm getting confused about terminology.
| |   + 22508 [wycats@gm il] I addressed some of this in the original proposal. Using a namespace would
| |   | 22515 [ola.bini@gm ] Yes, right. Sorry about that, I totally forgot that detail in your proposal.
| |   + 22512 [charles.nutt] Given that "Common List packages" is probably a fairly large topic, can
| |   + 22513 [jim.weirich@] Good call.  I understand "selector" to mean "message selector", i.e.
| |     22517 [charles.nutt] I regret not just doing this in the prototype, but I would see the
| + 22509 [charles.nutt] And these are my opinions...
|   22516 [ola.bini@gm ] Thanks Charles,
+ 22567 [quixoticsyco] "use" one class in place of another, but just allows you to prefer a
threads.html
top