199-435

^ Mem leaks in rb_str_become?
0199 [m.rokos sh.c] I'm not sure is following code doesn't mem leaks...
0200 [matz ruby-la] OK.

^ Bug in enum.c?
0201 [m.rokos sh.c] there could be bug in enum_sort_by? (Jump to !!!!!!! below)
+ 0202 [nobu.nokada ] Yes, segfaulted and fixed by your patch.
+ 0203 [matz ruby-la] It is a bug.  Commit it please.  Thank you.
  0204 [m.rokos sh.c] Done. Thank you.

^ openbsd system call changes
0205 [jfh gettysgr] I'm assuming that this is the right place to send proposed patches for
+ 0206 [nobu.nokada ] Right.
| 0207 [jfh gettysgr] Pah! You're right.
| 0214 [kjana dm4lab] The patch around setpgrp() let me recall the bug of autoconf-2.52
| 0215 [jfh gettysgr] Aha.  I'm using 2.52.  I will upgrade.
+ 0210 [matz ruby-la] Could somebody confirm this patch, especially on OpenBSD and other
  0211 [sean ruby-la] Works for FreeBSD 5.0-CURRENT.  I'm re-sending the patch as an

^ RSTRING macro in array context
0208 [jfh gettysgr] Is this a typo?
0212 [matz ruby-la] Yes, thank you for the patch.

^ OpenSSL in the base system?
0209 [sean ruby-la] What are the odds of having Michal Rokos's OpenSSL included in the
0220 [m.rokos sh.c] There could be some problem with licences of some

^ Consistent class naming for TCPSocket...
0213 [sean ruby-la] Attached is a patch to fix consistency issues with TCPSocket and

^ [PATCH]  lib/mkfm.rb:  have_bin()/find_bin()
0216 [sean ruby-la] This could be me over looking something, but in mkmf.rb, there is no
0217 [sean ruby-la] Whoops, wrong patch (the other patch was for fix to a security error
0218 [nobu.nokada ] I guess it's better to use CONFIG["EXEEXT"] and
0219 [sean chitten] [resent]

^ Wiring up the Boehm GC to the Ruby interpreter
0221 [matthew byte] I tried wiring up the Boehm GC (downloaded a couple of days ago) to the
+ 0222 [jfh gettysgr] Looks like what you've done is to replace ruby's calls to malloc with a
| 0223 [mattbee soup] No, I took out the calls to the Ruby GC.  There's obviously still some
+ 0224 [matz ruby-la] gc.c allocates big chunks of memory using malloc() and chop them into

^ Re: [gclist] Wiring up the Boehm GC to the Ruby interpreter
0225 [matz ruby-la] ruby-core@ruby-lang.org

^ speed improvements during ruby_init
0227 [jfh gettysgr] Messing around with gprof after being agitated by a recent thread

^ Patch to stop TCPSocket.new blocking on DNS lookups
0228 [mattbee soup] I'm using Ruby for a multi-threaded (50 or 60 threads at once) web spider and
+ 0229 [matz ruby-la] How about using "Socket.do_not_reverse_lookup = true"?
| 0230 [mattbee soup] The delays I'm experiencing were due to (necessary) forward lookups as far as
+ 0231 [decoux moulo] Well, getaddrinfo_a() is available only for glibc >= 2.2.4, no ?
  0232 [mattbee soup] Yup, I believe it's a pretty new call.  But unless we roll a DNS resolver
  0233 [matz ruby-la] Then have you looked at resolv-replace.rb ?  It replaces name lookup

^ Re: [PATCH]  lib/mkmf.rb:  have_bin()/find_bin()
0234 [nobu.nokada ] Sorry to be late.
0235 [matz ruby-la] I feel like "bin" in have_bin and find_bin (BTW, why both?) is better
0323 [sean ruby-la] executable works for me, I can update the patch.  The reason for both

^ [PATCH] resolv.rb
0236 [sdalu nic.fr] - it fixes the TCP Requester, even if not directly used by the DNS class,

^ [PATCH] object.c ruby.h (fwd)
0237 [skaav gmx.ne] my name is Robert Skarwecki and this is my first contribution to the Ruby
0239 [gotoken notw] Indeed introducing Boolean may make is-a relation clear in the
0240 [Dave Pragmat] For what it's worth, I think this is an evil side of the assert
0246 [gotoken notw] Indeed assert() of Test::Unit::Assertions does not test true/false.
+ 0247 [Dave Pragmat] Hmm - I differ here. I'd expect assert() to work the same way as 'if'
| + 0248 [nobu.nokada ] How about "true?"?
| | 0249 [Dave Pragmat] Well, to me, a.true? would imply the same as a == true, which isn't
| | 0250 [nobu.nokada ] Hmmm, then a.t?.  It would be lispy...
| | + 0251 [Dave Pragmat] Well, I guess the _ideal_ would be to have nothing, and to then have
| | + 0252 [gotoken notw] Isn't it too hackish :)
| |   0253 [Dave Pragmat] If you call Regexp#match with a non-string argument, it calls that
| |   + 0255 [gotoken notw] Oops! Sorry, I misunderstood that `and' or `or' treat conditions as
| |   | 0268 [masaki.suket] Older version of RubyUnit's assert() behavior is same as Test::Unit's.
| |   | 0269 [Dave Pragmat] There's not much we can do if people call the incorrect method :)
| |   | 0288 [masaki.suket] Ok. I will change the behavior of assert of RubyUnit like as Test::Unit.
| |   | 0295 [nahi keynaut] -1 for this change but I can live with it.
| |   | 0304 [masaki.suket] If I add assert_true method which checks 'foo == true',
| |   | 0313 [nahi keynaut] Yes, I can.  At the next, does method 'assert' become
| |   | 0315 [masaki.suket] No, at least I never deprecate 'assert' method.
| |   | 0320 [nahi keynaut] It's slightly regrettable for me that it won't be deprecated.
| |   + 0260 [kjana dm4lab] That's like my old proposal on.... where? :-) Matz rejected that since
| + 0258 [nahi keynaut] +1 to gotoken.  assert is for tester.  When someone put on
+ 0262 [m.rokos sh.c] I think if the RTEST returns 0 for Qnil or Qfalse, so Qnil is
  + 0263 [Dave Pragmat] Although this is a pure approach, I think it would make programming in
  + 0266 [matz ruby-la] Parhaps.

^ Boolean class (Re: [PATCH] object.c ruby.h)
0238 [matz ruby-la] * use unified diff (diff -u) instead of plain old diff to create

^ Compiling ruby on SGI origins
0241 [W.L.Kleb lar] I just tried compiling ruby-1.6.7 on a couple SGI origins (mips-sgi-irix6.5),
0245 [matz ruby-la] You need to supply the directory where libtk8.0 is, by --with-tcl-lib=
0264 [W.L.Kleb lar] Shouldn't configure have complained about not finding tk lib before I
+ 0265 [W.L.Kleb lar] Anyway, I did this and now I'm getting the same error *as* on
+ 0267 [matz ruby-la] It does on other platforms.  I don't know why SGI is different.
  0274 [W.L.Kleb lar] It appears that gcc is not configured correctly on either SGI

^ unifying nil and false (Re: [PATCH] object.c ruby.h)
0242 [matz ruby-la] "nil" and "false" are treated differently everywhere.  Unifying them
0243 [Dave Pragmat] I know, I know...  It was just a wistful question... :)
0244 [matz ruby-la] How about going back to Ruby version 0.95?  ;-)

^ A couple of questions on writing extensions
0254 [mattbee soup] *) what does 'Died with exception wrong argument type Object (expected Data)'
+ 0256 [mattbee soup] A bit more information: if I put a "print key.type" statement before & after
+ 0257 [nobu.nokada ] Literally, you passed something else to the method expecting
  0259 [mattbee soup] You're right, I think it was a misunderstanding with the mark/free functions

^ FYI: Updated Boehm GC
0261 [rich infoeth] I know there were some activities a bit back on this...

^ Who'd be the best person to handle this?
0270 [Dave Pragmat] I just received the following from Jim Freeze <jfreeze@freebsdportal.com>
0271 [decoux moulo] Not me :-)

^ Questions about installing in a corporate WAN
0272 [jim freeze.o] I have been trying to explore various install options
0273 [decoux moulo] Well, if you want a statically linked ruby, you must compile it with
0275 [decoux moulo] A small precision : you don't have problems with $SAFE <= 3

^ [Patch] Tini mini Range cleanup
0276 [m.rokos sh.c] there is nearly nothing to say about. Just 1 LONG2FIX conv, 1
0277 [matz ruby-la] Commit them.  Thank you.
0278 [m.rokos sh.c] Done. Thank you.

^ A truth? patch  + benchmarks
0279 [chr_news gmx] This is a multi-part message in MIME format.
0280 [decoux moulo] Not really sure but the values < 64 reserved for T_MASK
0283 [chr_news gmx] Hm, what about using FL_USER7? More generally, is there any
0284 [matz ruby-la] Unfortunately, there's no bit available for all objects.  All flag
0285 [chr_news gmx] To bad ... but maybe we might see something like it in 2.0? (or
+ 0286 [decoux moulo] Well, personnally I've not found a difference between the macro and inline
| 0287 [chr_news gmx] inline
+ 0289 [matz ruby-la] If it is really efficient, I think we can reserve a bit ot two for
  0294 [chr_news gmx] One bit should be enough.  My tests indicate that for the current
  0296 [nobu.nokada ] It's true with gcc 2.95.3 under i686-linux.

^ [NothinkToDo] ThreadSafeRuby
0281 [m.rokos sh.c] today I get a little bored, so I started to play with POSIX
0282 [decoux moulo] be carefull with this and see  [ruby-talk:5956]

^ (none)
0290 [flori nixe.p] some methods of File::Stat return 0 if the underlying feature isn't

^ Kernel Thread implementation
0291 [mattbee soup] charset="us-ascii"
0292 [decoux moulo] Put it in RAA

^ reply for your message
0293 [matz ruby-la] Hmm,
0307 [rich infoeth] returned

^ [Cleanup] GC longjmp macros
0297 [m.rokos sh.c] I think that these lines are odd in gc.c
0300 [matz ruby-la] Why did you feel so?
0301 [m.rokos sh.c] Because they are not in use in gc.c? (At least I cannot see any
0302 [nobu.nokada ] gc.c:1166-1169
0303 [m.rokos sh.c] True. (I got lost in ARCH def on line 1080.)

^ Is there a bug in Matrix or Complex?
0298 [jim freeze.o] I found what I would call strange behavior with Matrix and Complex.
0299 [gotoken notw] Elements above are floating point numbers, not integers.  This odd

^ File::Stat#blksize etc. should return nil instead of 0
0305 [flori nixe.p] Ok, nobody seems to use those platforms (MS Windows, I guess) anyway. ;)
0306 [matz ruby-la] Sounds reasonable.  If no one stands against the idea, I will apply

^ Q: OSSL in std. distr?
0308 [m.rokos sh.c] I just wanted to ask what should I do to 'OpenSSL for Ruby'
0310 [decoux moulo] * why OpenSSL *must* be in the standard distribution ?
0311 [m.rokos sh.c] The status is little vierd. The parts that are in OSSL are
0312 [jfh cise.ufl] I am -- thanks for developing it!

^ Progress on incremental GC?
0309 [tjw omnigrou] I noticed some messages in the archive about updating Ruby to use the

^ resolv.rb on 1.6.7
0314 [Dave Pragmat] I'm not sure if this is a ruby-core issue or not, so I'll keep it brief.

^ Embeding Ruby in C++ code.
0316 [kmak csd.aut] #include <ruby.h>
0317 [nobu.nokada ] You forget "self".

^ C and Ruby...
0318 [kmak csd.aut] How can i add an instance variable in a ruby class object?
0319 [nobu.nokada ] rb_ivar_set(obj, rb_intern("@foo"), Qtrue);

^ [ToDo] Ruby module, Warn(ing) interface
0321 [m.rokos sh.c] I've been reading through ToDo document and stuff like Warn(ing)
0322 [nobu.nokada ] I guess it's better to be printf-like as C version.

^ Mswin32 build flags
0324 [chr_news gmx] the current mswin32 build unfortunately (globally;-)
0325 [usa osb.att.] Because there are more problems with /Og flag
0422 [chr_news gmx] Thanks  (and sorry for the late reply)

^ Implications of a #force_free method in Object?
0326 [mattbee soup] I'm wondering whether it would be a sensible idea to implement a #force_free
0327 [nobu.nokada ] It sounds very dangerous, and the method wouldn't be able to

^ [Cleanup?] Int vs Long
0328 [m.rokos sh.c] because I have nothing to do and I'm quite bored, I went through
+ 0332 [jfh cise.ufl] How's ThreadSafeRuby coming along?
| 0335 [m.rokos sh.c] My idea was to bring real threads to Ruby (POSIX, Windows, or SW
+ 0333 [ryand zenspi] Is any work being done to speed up certain aspects of ruby? I was
  0336 [matz ruby-la] IKEGAMI Daisuke is working on faster bignum multiplication using

^ [Cleanup?] Int vs Long #2
0329 [m.rokos sh.c] ... and Diff is in this one :-))
0330 [matz ruby-la] It looks like OK.  Please commit.
0331 [m.rokos sh.c] Done. Thanks.

^ optimizing efforts?
0334 [ryand zenspi] yeah... i'm a dope. I need more coffee today. Resent w/ a real subject.

^ [Cleanup] Int vs Long (2nd part)
0337 [m.rokos sh.c] I just noticed that I forgot to change 1 int to long in
+ 0338 [nobu.nokada ] This changes semantics.
| 0339 [m.rokos sh.c] thanks for review.
| + 0341 [m.rokos sh.c] is this OK then?
| + 0343 [matz ruby-la] I think so too.  The current bahavior creates an hash even when the
+ 0342 [matz ruby-la] Commit this, please.
  0347 [m.rokos sh.c] Done. (Commited the patch without changing ary_new() to

^ [Cleanup] Int vs Long #3
0340 [m.rokos sh.c] this is the 3rd part of int x long cleanups.
0344 [nobu.nokada ] Although I doubt if they can exceed INT limit, well, no
0348 [m.rokos sh.c] thank you for a kind review again.
0350 [nobu.nokada ] $ diff --help | grep -we -p
0351 [m.rokos sh.c] Ah, grep - I didn't get the context... :-)

^ Bug in gc.c or misunderstanding on my part?
0345 [tjw omnigrou] I was just eyeballing through the Ruby source trying to evaluate it
0346 [cboos bct-te] rb_gc_stack_start is not only used for that, it is also

^ [Optim] Array#select
0349 [m.rokos sh.c] memory opt version of Array#select.
0354 [matz ruby-la] Accepted.
0357 [m.rokos sh.c] Commited.

^ Re: [Cleanup] Int vs Long #3 (RESEND with -p)
0352 [m.rokos sh.c] Here's the -p version.
0371 [m.rokos sh.c] I'm sorry, I didn't receive any feedback for ruby-core:352 (this
0374 [matz ruby-la] It's OK.  If you believe your fixes are mere cleanup, you don't need
0377 [m.rokos sh.c] Thank you, but I'd rather post the patches to ML first.

^ [Cleanup?] File (struct stat handling)
0353 [m.rokos sh.c] when I tried to track down all remaining off_t leftovers, I saw
+ 0356 [nobu.nokada ] Since File::Stat.allocate doesn't initialize DATA_PTR, this
+ 0359 [matz ruby-la] Why should we use memcpy instead of struct assigmnent?
  0370 [m.rokos sh.c] Now, I see what's wrong on my patch. Thanks nobu, matz.
  0375 [matz ruby-la] Could you explain why you prefer memcpy?  I'm just curious.
  0379 [m.rokos sh.c] Shame on me...

^ [Useless] Array and .*ALLOC.* changes
0355 [m.rokos sh.c] In ruby-core:33 I sent some useless patches to array.c. In
0360 [matz ruby-la] Go ahead.
0368 [m.rokos sh.c] Done.

^ [-Wall] node.h for eval.c
0358 [m.rokos sh.c] I touched node.h to make eval.c -Wall compilation happier.
+ 0361 [matz ruby-la] Accepted.
| + 0362 [nobu.nokada ] Sorry, it's unnecessary, since rb_eval() always restores
| + 0369 [m.rokos sh.c] Commited. Thank you.
+ 0365 [ryand zenspi] I just spent an LAX to SEA plane trip cleaning all but two warnings for

^ Ruby embedding limitations
0363 [tjw omnigrou] There are a couple of basic problems with Ruby's embedding support
0364 [vjoel PATH.B] Fork a subprocess to run ruby on each game type, and use drb to
0366 [ryand zenspi] I don't think this is a very satisfactory answer. We should admit that

^ RE: Welcome to our (ruby-core ML)         You are added automatically
0367 [girishs wrox] My email-id girishs@wrox.com may not be valid from next month on, could

^ [Safer] rb_class_path
0372 [m.rokos sh.c] about month ago I prepared this cleanup, but forgot it :-)
0373 [nobu.nokada ] +	str = rb_str_new(0, 2 + strlen(s) + 3 + SIZEOF_LONG * 2 + 1 + 1);
0378 [m.rokos sh.c] +	long len;
0380 [nobu.nokada ] It's better.
0381 [m.rokos sh.c] If you agree, I will commit this version. (It's based on my 1st
0383 [nobu.nokada ] I agree, and guess your previous patch has no problem.
0384 [m.rokos sh.c] commited.

^ Virtual host and port number in CGI
0376 [raymond theb] I am trying to figure out how to get the hostname and port # that was

^ [Update] Port match to new dup, clone framework
0382 [m.rokos sh.c] I spotted that match (in re.c) is not updated to new clone, dup
0385 [nobu.nokada ] You mean this? ;-)
0386 [m.rokos sh.c] I knew that you will definitely find some bug in it. (Shame on
0389 [matz ruby-la] OK if you add class check to become method (orig may not be MatchData),
0391 [m.rokos sh.c] Added and commited.

^ [Clenup?] rb_str_replace
0387 [m.rokos sh.c] condition FL_TEST(str2, ELTS_SHARED|STR_ASSOC) is allway true,
0390 [matz ruby-la] Thank you for finding.  Probably we need this too.

^ Native Threads (was Re: [Cleanup?] Int vs Long)
0388 [jfh cise.ufl] Hmmm...if it were possible to have a hash that atomically raised an

^ [Patch] io.c read_all
0392 [m.rokos sh.c] ===========================ChangeLog===============================

^ [MemLeak] in dln.c
0393 [m.rokos sh.c] am I right?
+ 0397 [matz ruby-la] Yes.  Thank you for finding a memory leak.
+ 0398 [nobu.nokada ] This looks tiresome.
  + 0402 [m.rokos sh.c] Of course it's ugly, but ... it was friday evening and I had to
  + 0403 [m.rokos sh.c] Well, personally, I'd add (before line with /* FIXME: Mem
    0404 [matz ruby-la] I'd use alloca.  I want dln.c to be separatable from Ruby as much as
    0405 [m.rokos sh.c] So, #2 is a winner.
    + 0406 [matz ruby-la] It seems OK (for me).  If you check if it works, go ahead to commit.
    | 0435 [m.rokos sh.c] Commited. (Including fix from ruby-core:407)
    + 0407 [nobu.nokada ] Ruby uses missing/alloca.c for such platforms, however I've
    | + 0409 [m.rokos sh.c] yes, I definitely missed this.
    | + 0411 [m.rokos sh.c] personally, I'd go for #3, than #1, and then #2.
    |   0418 [matz ruby-la] Since I want dln.c to be separatable, I prefer #2, #1, then #3.
    + 0408 [m.rokos sh.c] you have just introduced MEM leak, dude! :-)