1030-1293

^ Re: [ruby-cvs] ruby/ext/syck: ext/syck/handler.c (syck_hdlr_add_alias): add cast to avoid warning.
1030 [ruby-core wh] Thank you, eban.
1034 [matz ruby-la] How about forgetting 1.6.x, since it will be obsolete once we've

^ Kill support on Win32
1036 [djberge qwes] I would like to see Process.kill support added for Win32.  I've written a
1038 [usa osb.att.] Ruby 1.8 already has `kill' support.
1046 [djberge qwes] Actually, I didn't see this before, but I think my approach still has

^ ping.rb
1037 [dave pragpro] On my Mac OSX,
+ 1039 [ryand zenspi] <497> telnet localhost echo
| 1040 [dave pragpro] Yes - but 'connection refused' should generate ECONNREFUSED, which is
| 1042 [nobu.nokada ] Agree, but I don't have Mac OS X box.  Can't you get strace
| + 1043 [dave pragpro] ktrace ruby -r socket -e "TCPSocket.new('localhost', 'echo')"
| | 1044 [nobu.nokada ] Seems here should be the problem.
| | 1045 [dave pragpro] I can do this, but I need to know what to test for to see if I'm on an
| | 1047 [dave pragpro] I just tried using the Cygwin code under OSX, and it doesn't seem to fix
| | + 1048 [djberge qwes] I believe __APPLE__ is a gcc defined macro.  I dunno if CodeWarrior or
| | | 1050 [nobu.nokada ] According to http://predef.sourceforge.net/index.php,
| | | 1054 [eric FreeBSD] I think that is wrong. I've used just __APPLE__ in the past. Here's some
| | | 1055 [nobu.nokada ] Thank you.  Hmmm, would you send the correction to the
| | + 1049 [nobu.nokada ] Mmmm, how do other programs (telnet, ping ...) handle that
| |   1051 [dave pragpro] I'm guessing the problem is caused by the socket being opened in nowait
| |   1053 [nobu.nokada ] Although it's ad hoc, what about treating EINVAL preceded by
| |   1058 [dave pragpro] It seemed to work: many thanks!
| |   1063 [jlst gettysg] I think that OSX and cygwin are separate cases.  Maybe delay could be
| |   1064 [jlst gettysg] No I'm wrong. I was confused. Your patch works like charm.
| + 1052 [ryand zenspi] Come hang out and give a talk at Seattle.rb and I'll get you as much
|   1056 [nobu.nokada ] Thank your very much for the proposal, but Seattle is so far a
+ 1041 [matz ruby-la] * OS X connect(2) behave differently

^ Fwd: ping.rb
1057 [jlst gettysg] __MACOSX__ does not seem to be set but __APPLE__ is.
1059 [nobu.nokada ] Sounds great.
1060 [matz ruby-la] Could you commit the fix?
1061 [nobu.nokada ] Although I'm not sure if this works on cygwin, OK?  The box I'm
1062 [matz ruby-la] It's OK, since no one notice the problem unless you commit.

^ [1.8] allocate
1065 [decoux moulo] ruby don't like ::allocate

^ YAML problem
1066 [akr m17n.org] I found a problem with yaml.
1067 [nahi keynaut] Not related to original problem but Syck should raise
1068 [ruby-core wh] Thank you.  This is resolved.  Still working on some encoding issues
+ 1069 [akr m17n.org] % ruby -ryaml -e 's = {"1"=>2}.to_yaml; puts s; p YAML.load(s)'
| 1071 [akr m17n.org] It's fixed.  Thanks.
| 1120 [ruby-core wh] Fixed.  Was a problem with anchors on a collection.
| + 1121 [nobu.nokada ] Seems that porinters to local variables would be stored in
| | 1133 [matz ruby-la] Hmm, it seems so.  I will remove the variable.
| + 1134 [akr m17n.org] Thank you.
+ 1070 [matz ruby-la] I've believed I had a logic in choosing TypeError and ArgumentError,
+ 1072 [nahi keynaut] Don't take that too much.  YAML.rb should not depend Ruby's
  1073 [ruby-core wh] I just want YAML.rb to be on its best behavior with Ruby.  If a User has

^ crash involving Qundef
1074 [matju sympat] that only uses "xmlparser" (libexpat.so) to read its input file, and then
1075 [matz ruby-la] Qundef is the value for internal use that should not reveal to
1076 [nobu.nokada ] Just download moulinette.rb and following XML files.

^ [1.8] terminated object / invalid inspect_tbl
1077 [decoux moulo] Normal these messages ?
1078 [nobu.nokada ] Must not be.
1079 [decoux moulo] There is also "invalid inspect_tbl" in array.c, no ?
1080 [nobu.nokada ] It is fixed already in array.c 1.121.
1081 [decoux moulo] Well, it work with hook (I must stop to write stupid thing :-)))
1085 [nobu.nokada ] Have you implemented method hook?
1089 [matz ruby-la] Yeah, I'd like to see the code.
+ 1090 [nobu.nokada ] It also seems to fix [ruby-core:01074].
+ 1092 [decoux moulo] It's a toy actually : it don't work yet with alias and it need some other
  1093 [decoux moulo] a better version : it must work with operator, class local variable

^ Possible problem with mkmf under Darwin
1082 [dave pragpro] I've been having fun getting readline support to work on OSX. My latest
+ 1083 [decoux moulo] There is a missing `()' which can explain the error message
+ 1084 [nobu.nokada ] Before it, isn't there this test?
| 1087 [dave pragpro] Yes, there was... but why would it try the second one if the first one
| 1088 [nobu.nokada ] The second would not be done if the first were succeed.
| 1091 [dave pragpro] No, but now that I know what to look for I'll be keeping my eyes open.
+ 1086 [jim freeze.o] /System/Library/PrivateFrameworks/readline.framework/readline

^ Re: [ruby-cvs] ruby, ruby/lib: * eval.c (ev_const_defined, ev_const_get), variable.c
1094 [dave pragpro] This is probably a silly question, but does autoload ever get used? Or
1095 [nobu.nokada ] Kernel is a Module, so still you can call autoload everywhere.
+ 1096 [dave pragpro] used in the real world, or should we be deprecating autoload, and
| 1098 [ged FaerieMU] I would have been using it for years now if it worked the way the
+ 1097 [dave pragpro] I have to say that this behavior is somewhat surprising. From its name,
  1101 [nobu.nokada ] Thank you.  Fixed now.
  1102 [matz ruby-la] I think autoload should require X::XXX to be defined, so that should
  1104 [nobu.nokada ] At first I thought so too.  But, in this case, refereced is
  1105 [matz ruby-la] ??
  + 1106 [nobu.nokada ] Yes, it should be X::XXX now.
  + 1109 [dave pragpro] But either way, shouldn't autoload be compatible with load? After all,
    1111 [matz ruby-la] I consider my proposal compatible.  It "automatically loads" the file

^ Isn't NUM2UINT broken for 64-bit?
1099 [lyle knology] All,
1100 [nobu.nokada ] I guess it's a bug.
1103 [matz ruby-la] Commit the fix, please.

^ large file compilation issue on solaris with 1.8
1107 [djberge qwes] gcc 3.2
1108 [holmberg iar] I think Ruby has started to support large files since 1.6.8.

^ [1.8] alias
1110 [decoux moulo] I still don't understand alias, normal this ?
1112 [matz ruby-la] No.  It's a bug.  Will be fixed soon.
1113 [decoux moulo] Normal ?
1114 [matz ruby-la] No, I was too eager to resolve inheritance line problem.  I will move
1116 [decoux moulo] Well, I like the DRY principle but certainly I don't really understand how
1117 [matz ruby-la] Eban also mentioned a bug in my fix.  I will cover it.

^ ext/gdbm and multiple readers at a time
1115 [batsman.geo ] man gdbm says

^ minor ruby-mode.el issue
1118 [rpav nwlink.] It seems ruby-mode uses ^ and $ in the regexps for beginning-of-defun

^ [1.8] $SAFE / exit
1119 [decoux moulo] Normal ?
1135 [matz ruby-la] $SAFE will be turned of during finalization.

^ [1.8] Exception::new
1122 [decoux moulo] Normal ?
1123 [nobu.nokada ] Contains fix for (ruby-bugs-ja:PR#487) and [ruby-core:01119].
1124 [decoux moulo] Work for this
1126 [nobu.nokada ] Hmmm, at such disaster, nothing might be possible.
1131 [nobu.nokada ] Sorry, this wasn't enough.
1136 [matz ruby-la] Commit this fix, please.

^ [OT] A question on input encodings
1125 [dave pragpro] Sorry - I know this is OT, but...
+ 1127 [nobu.nokada ] It is impossible to distinguish SJIS and UTF-8 by their data.
+ 1128 [austin halos] I'm not sure that this is the best way to do it, personally. I'd
+ 1132 [akr m17n.org] It is not practical.

^ RubyPKI and Ruby 1.6.8
1129 [rich infoeth] Does RubyPKI (ossl-0.2.0.pre2) run on Ruby 1.6.8 or is 1.8 required?
1130 [nahi keynaut] Requires 1.8 and OpenSSL/0.9.7.

^ Unexpected arity of method defined by Module#define_method
1137 [ged FaerieMU] Is this expected behavior?
1138 [decoux moulo] 1.6.8 and you'll see the difference with 1.8
1139 [matz ruby-la] Thank you for fixing a bug.

^ Rubicon questions
1140 [holmberg iar] I just tried to run Rubicon on the latest version of Ruby
1141 [chadfowler c] - Is the latest CVS-versions of Rubicon and Ruby supposed to "agree" ?
1142 [dave pragpro] Rubicon was an attempt to give the Ruby committees a way to test
+ 1143 [matz ruby-la] I (among others) will fix after releasing 1.8.0.  Then I propose
+ 1144 [holmberg iar] That seems like a very good idea.
  1145 [dave pragpro] If you look through the existing tests you should find examples of

^ File.link error messages ?
1146 [holmberg iar] File.link(from,to) can give an error for a number of reasons.

^ Copying RVALUE
1147 [ruby-core wh] Without going into terrible details, I am looking for a way to move an
1155 [matz ruby-la] Each object does not assume bitwise copy.  It may cause problems, such
1157 [ruby-core wh] In YAML, references are kept through an ANCHOR/ALIAS system.  Consider
1160 [akr m17n.org] One idea is creating empty array just after the anchor.
1161 [ruby-core wh] This is an idea, though I'd like something more general.  The anchor
1162 [ryand-ruby z] Couldn't you use a one element array as a pointer-pointer? I assumed
+ 1163 [matju sympat] There's something called Object#become in Ruby 1.8 but I don't quite get
| + 1164 [nobu.nokada ] It had been renamed as "initialize_copy", and never change the
| + 1170 [decoux moulo] because it was too dangerous ...
|   1171 [matju sympat] 1. what does "dangerous" means to you, and how would i tell between such
|   1172 [decoux moulo] It was easy to crash any script written in ruby
|   1192 [matju sympat] It's already easy to crash anything. Quite every powerful feature can be
|   1195 [decoux moulo] Just one question : are you a windows or a linux user ?
+ 1165 [akr m17n.org] No.  My recommendation is "object allocation should be done before it
  + 1166 [ruby-core wh] Yes, exactly.  You describe it succinctly and with true examples, Akira.
  | + 1167 [akr m17n.org] In this example the parser can know the type just after reading '['.
  | + 1169 [vjoel PATH.B] What about a delegation approach? Or is that too inefficient?
  + 1168 [ryand zenspi] This exact discussion is going on in the little languages mailing list.

^ 'unexpected break' when captured block calls break
1148 [george.marro] Proc-closures capture any block passed to their enclosing scope: the script
1156 [matz ruby-la] This is a bug.  Thank you for finding it.  I suspend preview3 release
1158 [nobu.nokada ] Although I'm not sure what will be expected, kinds like this?
1159 [matz ruby-la] Yes, it's almost same with my fix.  But I forgot putting volatile to

^ stat/chmod on Windows
1149 [holmberg iar] When trying to understand the cause I realized that the

^ Bus error in compiler
1150 [dave pragpro] $ cat t.rb
1151 [decoux moulo] Can you try ?
1152 [dave pragpro] $6);

^ YAML parse error with YAML.load("\n\0".to_yaml)
1153 [akr m17n.org] YAML raises parse error as follows.
1154 [ruby-core wh] Akira, thank you once again.  Often in the afternoon when I am listless,

^ [Patch] class.c code cleanup (rb_class_*_instance_methods)
1173 [jivera flame] introducing new patches (I couldn't find anything about it on the
1174 [nobu.nokada ] This will let all methods be warned as "instance_methods".
1175 [jivera flame] Er, sorry, I guess I missed that. :/
1176 [nobu.nokada ] rb_id2name(rb_frame_last_func())
1193 [jivera flame] I actually used "as of 1.8.1" because that seemed the clearest to me
1194 [nobu.nokada ] Although I'm not an English speaker and am not sure, it sounds
1196 [jivera flame] Eh... It figures that I'd miss something like this... okay, but if
1197 [matz ruby-la] It's already in.
1198 [nobu.nokada ] Warnings are emitted.
+ 1199 [jivera flame] Ugh, what compile flags did you set to get these warnings?  They were
| + 1200 [jivera flame] Ack, the recurse = Qnil line should acually be recurse = Qfalse... patch
| + 1201 [nobu.nokada ] It depends on gcc version.
|   1202 [jivera flame] That's odd, I have the same specs...
|   1203 [nobu.nokada ] Hmmm, sorry, it may be caused by inlining with -O3.
+ 1204 [matz ruby-la] Thank you.  Commit the fix, please.

^ Re: In 1.8.0 nil.to_s is not the same as ""
1177 [decoux moulo] [[ followup on ruby-core ]]
1178 [nobu.nokada ] Should it occur same exception as Dir.open("")?
1179 [decoux moulo] The problem is not here : do someone has verified *all* the use of
+ 1180 [decoux moulo] See
| + 1183 [decoux moulo] svg% ruby -e 'load String.allocate'
| + 1184 [decoux moulo] Well, I'll stop to post all bugs one by one :-(((
+ 1181 [nobu.nokada ] You mean that ptr should be guaranteed not to be NULL?
| 1182 [decoux moulo] It must be tested otherwise ruby segfault in some case
+ 1185 [matz ruby-la] I assumed tests should be done before releasing 1.8 when I changed ptr
  + 1186 [decoux moulo] Probably best because actually I've (which give a segfault)
  + 1187 [matju sympat] If you want to save memory, then instead of pointing to NULL, it could be
  | 1189 [matz ruby-la] Thank you.  Actually, it's exactly what I did.
  + 1188 [nobu.nokada ] Well, all I found are attached.
    1190 [matz ruby-la] Incredible speed.  After my change to use constant empty string, some

^ Standard type conversion mechanism
1191 [rpav nwlink.] OK, I'm posting this because a few people on #ruby-lang thought it

^ class_eval, private
1205 [matju sympat] class Foo; end; Foo.class_eval {def hello; end}; Foo.new.hello
1206 [matz ruby-la] No.  It's a bug to be fixed.

^ warning in ruby extension eats memory
1207 [Eugene.Scrip] This message was posted to ruby-talk, but I didn't get responce from
1208 [decoux moulo] Well, I was not able to reproduce the problem with
1209 [Eugene.Scrip] I don't see warning, script just halts consuming all the memory and
1210 [decoux moulo] Something like this
1211 [Eugene.Scrip] OK, I've updated my ruby to 2003-07-02 and have done what you suggested.
1212 [decoux moulo] Really I don't see. When it crash can you do ?
1213 [Eugene.Scrip] (gdb) up 3
1214 [decoux moulo] Yes, and because it has flags == 0, it call #write and enter in an
1215 [Eugene.Scrip] The question is: why I experience this problem but others do not, and
+ 1216 [decoux moulo] This is the beauty of the GC :-)))
| + 1218 [Eugene.Scrip] That is the question! You are ruby hacker, not me :)
| | 1220 [decoux moulo] What is your value of ulimit (for stack_size) ?
| + 1221 [nobu.nokada ] rb_call() checks only receiver's class, but nothing for flags.
|   1222 [decoux moulo] Yes, but what I don't understand is how ruby can have all these calls
|   1224 [nobu.nokada ] I can't get what you mean, these seem to be called normally.
|   1226 [decoux moulo] 108455 call ?
|   1227 [nobu.nokada ] I see.  Didn't think you mean the level.
+ 1217 [nobu.nokada ] This is strange; rb_deferr is hooked by $deferr and should
  + 1219 [decoux moulo] It's called at the end of ruby : don't forget that test-unit is special.
  | 1223 [nobu.nokada ] Sorry, I missed it.  You're right.
  | 1225 [matz ruby-la] And my final words should be "don't do anything non trivial in free
  + 1228 [Eugene.Scrip] I'm sorry for late response.

^ stack problem
1229 [matju sympat] If I embed Ruby into certain programs, I have this problem: when
1230 [batsman.geo ] You can use Init_stack, see the thread starting at [ruby-talk:74400].
1231 [matju sympat] Thanks. If I understood correctly, I have to use it like this?
1233 [matju sympat] As an explanation for the code snippet: I want to find the bottom of the
1234 [batsman.geo ] If you pass NULL Ruby will take the address of a variable in the stack,
1235 [matju sympat] you mean recording the top of the system stack, and using it as a bottom
1236 [matz ruby-la] It would be.  But I have no idea yet to accomplish.
1278 [matju sympat] This is my current cleaned-up code for finding the bottom of the stack.
1279 [nobu.nokada ] Stepping and rounding by pagesize might be better.
1280 [matju sympat] I don't know what makes you think that rounding by pagesize might be
1281 [nobu.nokada ] Just it would be faster.  No matter.
1282 [matju sympat] Ok, for the speed improvement, I'm not convinced, unless someone finds a
1283 [nobu.nokada ] Agree.  It's just a thought.
1292 [matju sympat] Ok, sounds right, after all.
1293 [matz ruby-la] Last time I got complained for old Sun compiler long time ago.