8175-12006 subjects 8569-9548

:'foo-bar' causing warnings in rdoc parser
8382 [flori@ni e. ] This would stop the warning of course, but it is actually valid Ruby code.

[ ruby-Patches-5177 ] Win32 NTLM/Negotiate Authentication Support for Net::HTTP
8383 [noreply@ru y] Patches item #5177, was opened at 2006-07-24 08:28

ri patch for html format
8392 [ksruby@gm il] ri doesn't escape <,>, &, etc. characters when used with html formatter like
8679 [ksruby@gm il] Anyone?
8730 [drbrain@se m] I checked in a modified patch to HEAD on August 4th and just now to

.document patch
8412 [hgs@dm .a .u] The intent of the patch below is to improve the coverage of the rdoc

[ ruby-Bugs-5203 ] date.rb: DateTime.sec_fraction and day_fraction_to_time return incorrect second fraction
8413 [noreply@ru y] Bugs item #5203, was opened at 2006-07-26 14:15
8417 [matz@ru y- a] Can you show us a script to reproduce the problem?  It seems like no

[ ruby-Bugs-5217 ] irb: assignment to $stdout
8419 [noreply@ru y] Bugs item #5217, was opened at 2006-07-27 16:12
8421 [nobu@ru y- a] It was an issue of ext/readline, and TypeError raises on recent

Logger -- documentation patch
8422 [hgs@dm .a .u] A patch against

doc patch: readbytes.rb
8423 [hgs@dm .a .u] A patch against the stable snapshot.
8424 [Daniel.Berge] Minor tweak: "Reads exactly +n+ bytes from the IO stream ..."
+ 8425 [Daniel.Berge] Shift those one character to the left.  Damned variable width fonts...
+ 8426 [hgs@dm .a .u] Yes, agreed.  (The carets showed up in the right place for me, as
  8431 [drbrain@se m] I've been busy and haven't had time to verify/check in your changes.
  8462 [hgs@dm .a .u] OK, thank you.  I know it's still on the radar now. :-)

RDoc picking up comments from function prototypes
8427 [tilman@co e-] #ifdef FOO
8574 [tilman@co e-] Of course nobody does, so I did it myself.
+ 8576 [drbrain@se m] Unless I get approval from matz it is too late for 1.8.5.
| 8586 [tilman@co e-] Seems like my timing was "perfect" for this one :(
+ 8601 [akr@fs j. rg] Tha patch makes an warning: "No definition for sock_accept_nonblock"
  8602 [tilman@co e-] T081315.diff.txt.gz

doc patch: weakref.
8428 [hgs@dm .a .u] Here is a doc patch for weakref, against the stable snapshot as before.
8429 [pit@ca it in] Hugh, thanks a lot for your ongoing work on Ruby's documentation!
8430 [Daniel.Berge] Something about this phrase bothers me.  How about simply, "A weak
8434 [matz@ru y- a] But it's not true.  WeakRef class or any other class does not perform
8436 [djberg96@gm ] Then your description (slightly tweaked) is what we want. :)
+ 8437 [ mfp@ac .o g] Some people don't know how Ruby's GC works, or even what a mark&sweep GC is.
| + 8451 [matju@ar en ] ...
| + 8463 [hgs@dm .a .u] OK, how is this?
|   8467 [drbrain@se m] I think this is good enough.
|   8470 [hgs@dm .a .u] I just thought something in the introductory text would provide context
|   8481 [drbrain@se m] It also ends up being documented twice.
|   8487 [hgs@dm .a .u] OK, violation of DRY in comments is probably as bad as code for maintennce.
|   + 8488 [meta@po ox c] I strongly disagree. The purpose of library API documentation is to act
|   | 8490 [hgs@dm .a .u] OK, you've convinced me that this leads to broken encapsulation.
|   | 8491 [meta@po ox c] I've explained before that unit tests are not a *specification* of
|   | 8492 [hgs@dm .a .u] They express intent in a machine-useable way.  They may be incomplete,
|   | 8519 [meta@po ox c] Yes, but it's not the machine that cares what the API is. APIs are for
|   + 8495 [drbrain@se m] I like it.
|     8500 [matz@ru y- a] Eric, you can commit the doc fixes that you like.  But since 1.8.5 is
+ 8438 [matz@ru y- a] If our documentation maintainers feel it right, I have no objection,

Problem with ruby 1.9
8432 [vincent.isam] Recently I've had some problems with ruby 1.9. It's at least 2 weeks
8433 [akr@fs j. rg] I think it is a gcc's alloca problem.
8435 [vincent.isam] I tried it and indead it works fine with ./configure CFLAGS='-O2 -g -DC_ALLOCA'.

Inconsistency in scoping during module_eval?
8441 [headius@he d] ...
8442 [nobu@ru y- a] Nobu Nakada
8443 [headius@he d] ...
+ 8444 [nobu@ru y- a] Constants definitions/lookups are decided at compile time.
+ 8445 [matz@ru y- a] Constant assignments follows static scoping in 1.8.  You have to use
  + 8446 [chiology@gm ] Does it matter that you're using module_eval inside of a class definition?
  | 8450 [matju@ar en ] ...
  + 8454 [headius@he d] ...
    + 8455 [chiology@gm ] I don't want to take this thread off-course, but what I meant was
    | 8458 [matju@ar en ] ...
    | 8459 [headius@he d] ...
    | 8460 [chiology@gm ] Awesome. Thank you very much for your responses. Curious. I knew a
    | 8461 [jan.svitok@g] I think it's the oher way round ;-)
    | 8468 [chiology@gm ] Yes, absolutely. However, if everything is an object (including a
    | + 8469 [matju@ar en ] Each module and each class have a class which are all subclasses of the
    | + 8471 [dblack@wo bl] class Class < Module
    |   8472 [chiology@gm ] I think I'll just take an afternoon (or 20) and experiment until I
    + 8456 [matju@ar en ] ...

[ ruby-Bugs-5243 ] #if should be #ifdef in ruby.h
8447 [noreply@ru y] Bugs item #5243, was opened at 2006-07-30 16:31
8448 [shyouhei@ic ] Why on earth?
8449 [now@bi wi se] Because if LONG_LONG_VALUE /isn't/ defined, it'll cause a compiler
8452 [shyouhei@ic ] I believe ANSI C defines that an undefined macros showld be expanded to
8464 [now@bi wi se] Any identifiers remaining after macro expansion are replaced by 0L.
8465 [shyouhei@ic ] preference, not a matter of it should or should not.

[ ruby-Patches-5244 ] RingServer fails to find a TupleSpace when TupleSpace resides in the same ruby process with RingServer
8453 [noreply@ru y] Patches item #5244, was opened at 2006-07-31 00:17

bypassing method lookup/call overhead
8457 [eric.mahurin] From performance testing that I've looked at with a C extension, it

Multi-Line Date Formate Patch
8466 [james@gr yp ] require "date"
8474 [james@gr yp ] Uh, hello?  Is there any good reason not to apply this patch?
8475 [nobu@ru y- a] It seems insufficient.
8476 [james@gr yp ] Ah, I did miss the method in Date.  Good catch.  However, I doubt we
+ 8482 [nobu@ru y- a] Indeed.
| 8483 [james@gr yp ] Great have we reached the point where we can safely apply a patch then?
+ 8508 [james@gr yp ] Just following up here...
  8509 [matz@ru y- a] If Nobu is Ok for the fix, he can check in the patch.  I think
  8510 [nobu@ru y- a] I'm OK.
  8511 [james@gr yp ] You are welcome to include my address, sure.

ping.rb doc patch
8473 [hgs@dm .a .u] A patch to move the module docs closer to the module, add a comment for

bug in Range
8477 [neleai@se na] I tried my program with 1.9 but there was another bug with char handling. Bug is in this

resolv.rb -- doc patch.
8478 [hgs@dm .a .u] This is an attempt to get the RD format docs for resolv.rb into
8480 [drbrain@se m] HEAD already has an RDoc'd resolv.rb.
+ 8486 [hgs@dm .a .u] OK, I'll look into that later.  I've been working from the tarballs.o
+ 8494 [hgs@dm .a .u] Yes.  I think that's better than mine.  Thank you.

Re: [Correction] resolv.rb -- doc patch.
8479 [hgs@dm .a .u] Just spotted a typo in that patch which affects code.

[ ruby-Bugs-5263 ] strptime fails to properly parse certain inputs
8484 [noreply@ru y] Bugs item #5263, was opened at 2006-08-01 23:14
8485 [matz@ru y- a] Just because strptime() %Y parsed 20060401 as a year of 20,060,401
8538 [nobu@ru y- a] How about to accept an optional width?
+ 8539 [znmeb@ce ma ] The name "strptime" is used in other programming languages for this
| 8540 [shyouhei@ic ] Why bother other languages?  They are on their own.  We should not
| 8542 [ryand-ruby@z] *kiss*
| 8543 [matju@ar en ] ...
| + 8544 [znmeb@ce ma ] Yeah ... even ARC is only supposed to last 100 years :)
| + 8549 [meta@po ox c] And presumably not an astronomer. You don't have to be immortal to have
|   8551 [matju@ar en ] ...
|   8553 [matz@ru y- a] It does.
+ 8545 [matz@ru y- a] I am not sure what you mean by 'optional width'.  Sorry for being too
  8548 [nobu@ru y- a] Maximum width to be converted, like as scanf(3).

[ ruby-Bugs-5267 ] exception loss
8489 [noreply@ru y] Bugs item #5267, was opened at 2006-08-02 12:00
8554 [matz@ru y- a] I am not sure what caused a problem and what you have done to fix it.

Re: doc patch: weakref.rb - against HEAD
8493 [hgs@dm .a .u] Here is a patch against HEAD from CVS grabbed this afternoon,

[ ruby-Bugs-5273 ] range ===
8496 [noreply@ru y] Bugs item #5273, was opened at 2006-08-02 21:49
8501 [matz@ru y- a] Right, it calls Object#=== which should be Enumerable#member?
8503 [neleai@se na] ...
8506 [nobu@ru y- a] It might make a sense.

Ruby Socket to support SCTP?
8497 [philippelang] Is there a plan to support SCTP Sockets in Ruby?
+ 8498 [aredridel@nb] I've been thinking about implementing it in my copious free time for a
| 8499 [garbagecat10] Want help?
+ 8520 [sroberts@un ] If the one-to-one API described in
  8521 [philippelang] You're right, one-top-one should work with roughly the same calls as TCP

doc patch for mailread.rb
8502 [hgs@dm .a .u] Here is a documentation patch for mailread.rb.  (HEAD and stable seem

TCPSocket: bind method missing
8504 [hadmut@da is] I just tried to write an rsh client in Ruby.
8505 [garbagecat10] Have you tried using the Socket class? You have to set your own
8507 [hadmut@da is] Yup, works. Although some overhead for struct sockaddr.

[#5203] date.rb: DateTime.sec_fraction and day_fraction_to_time return incorrect second fraction
8512 [ruby-core@es] I'm following up to the mailing list directly since Tom Copeland advised me
8552 [matz@ru y- a] I asked the author, and he told me it was intentional (being ratio to

patches for the 1.8.5 deadline...
8513 [hgs@dm .a .u] As far as I can tell the only patches which I've submitted which
8515 [drbrain@se m] I have time tonight and tomorrow night and sunday to work on these.
+ 8516 [headius@he d] ...
| 8518 [matz@ru y- a] For 1.8.5?  No.
+ 8517 [matz@ru y- a] Thank you.
+ 8523 [hgs@dm .a .u] Thank you.  It may be worth considering other .document files as
  8528 [drbrain@se m] Ok, I have checked in to both HEAD and 1_8.  Please review for any
  8547 [hgs@dm .a .u] Thank you.  That all looks to be as I wished.  It looks like you
  8550 [drbrain@se m] Curious.  For some reason the .document file never merges correctly,
  8556 [hgs@dm .a .u] No, by "I've not seen" I meant "areas of the source I've never got

[ ruby-Bugs-5293 ] "ri File.open" is misleading
8514 [noreply@ru y] Bugs item #5293, was opened at 2006-08-03 11:38
8534 [matju@ar en ] ...

[PATCH] IRB change for RDoc workaround
8522 [drbrain@se m] ...

[PATCH] readline 
8524 [neleai@se na] ...
8558 [matz@ru y- a] I don't think the names keyseq and getc_proc reflects their role
8559 [nobu@ru y- a] * no boundary check for the Array
8592 [neleai@se na] That array is for internal use. Unless you type record separator it doesnt crash.
8600 [nobu@ru y- a] Then, you should not set it to an instance variable.
8615 [neleai@se na] I rewritten my patch

rdoc bug?
8525 [steven.jenki] I think I've found a bug in rdoc's handling of C files. Specifically, it
+ 8526 [drbrain@se m] Can you give a concrete example of where this happens?
| 8527 [steven.jenki] #include "ruby.h"
+ 8529 [neleai@se na] Bugs are reported via rubyforge. I wonder why there isn't link at ruby-lang.org
| 8531 [steven.jenki] Thanks.
| 8533 [steven.jenki] OK, I see now. The Ruby project, not the rdoc project.
+ 8530 [steven.jenki] @@ -516,7 +516,8 @@

[ ruby-Patches-5310 ] improved debugger
8532 [noreply@ru y] Patches item #5310, was opened at 2006-08-05 11:03

sandbox r42 + swap struct BLOCK?
8535 [ruby-core@wh] considerate testing.
8537 [chiology@gm ] This is some really cool stuff, _Why. Thanks for the great work. If

[ ruby-Bugs-5319 ] rdoc documents C module methods as instance methods (patch)
8536 [noreply@ru y] Bugs item #5319, was opened at 2006-08-05 12:13

[ ruby-Bugs-5335 ] Ruby 1.8.4 compiled with profiling can't install the Rdoc files
8541 [noreply@ru y] Bugs item #5335, was opened at 2006-08-06 15:55

[ ruby-Bugs-5339 ] [PATCH] spurious "cross-thread error in rb_thread_schedule"
8546 [noreply@ru y] Bugs item #5339, was opened at 2006-08-07 09:08

[ ruby-Bugs-5351 ] Process.gid= fails on OS X
8555 [noreply@ru y] Bugs item #5351, was opened at 2006-08-08 01:56
+ 8557 [matz@ru y- a] Strange.  It's just calling setgid() from libc.  Can you show us your
+ 8560 [zn@mb .n ft ] is used in Mac OS X, and setregid(nonzero, -1) failed.
  8564 [nobu@ru y- a] The setregid() function is used to set the real and
  8565 [nobu@ru y- a] Oops,
  8587 [zn@mb .n ft ] I tried patch with debug print, s/OBSOLETE_SETREUID/BROKEN_SETREUID/g,
  8588 [nobu@ru y- a] Of course, not only setregid().
  8589 [zn@mb .n ft ] Mac-mini:~ kazu$ sudo ~/opt/ruby18/bin/ruby -e 'include Process::Sys' -e 'p setregid(502, getegid)'
  8590 [nobu@ru y- a] Okay.  Since it feels a show-stopper, I'd like to commit it
  + 8591 [zn@mb .n ft ] I thinks that a patched ruby on Mac OS X behaves
  + 8625 [matz@ru y- a] Can you commit?

sandbox timers & block scopes
8561 [ruby-core@wh] 1. Ruby uses setitimer to switch threads.  I would like to use setitimer as well
+ 8624 [ruby-core@wh] raise ThisDecayingInquisition, "anyone? anyone at all?"
| 8627 [mental@ry ia] e.
| 8628 [ruby-core@wh] Yeah, that's just it.  I don't want to kill the whole thread containing the
| + 8629 [headius@he d] ...
| | 8630 [headius@he d] ...
| | 8633 [ruby-core@wh] Okay, well, I'll try the watchdog thread and SIGALRM (my own timer) ideas.
| | 8636 [garbagecat10] ...
| | + 8638 [mental@ry ia] It's probably also worth asking how he's going to get SIGALRM to work on
| | | + 8640 [neleai@se na] ri Kernel#fork
| | | | 8645 [mental@ry ia] Elaborate?
| | | | 8655 [neleai@se na] Now is best way spawn new process by(but its slow)
| | | + 8641 [garbagecat10] I sortof decided not to go there. One-problem-at-a-time kind of thing.
| | + 8649 [ruby-core@wh] Normal Ruby threads use SIGVTALRM.  The SIGALRM would check all open sandboxes
| + 8631 [mental@ry ia] Well, I was thinking of injecting a timeout exception with
|   8632 [ruby-core@wh] Mmnnn, right right yah ok, I'll look into this then.  Maybe there's a way to
|   8639 [mental@ry ia] to=20
|   8646 [mental@ry ia] I've got one last thought on this:  maybe I'm looking at the problem the
+ 8642 [garbagecat10] I'm not sure if I'm about to repeat what Charles has already said, but
| + 8643 [Daniel.Berge] You don't.  You use events.
| | 8647 [mental@ry ia] That makes sense.  Out of idle curiosity, do you know if there's a
| | 8648 [Daniel.Berge] html/UCMGch09.asp
| + 8644 [headius@he d] ...
|   8650 [garbagecat10] Shouldn't take much to validate this approach- just dig into the
|   8651 [mental@ry ia] Hmm, if I understand you correctly, isn't that what rb_thread_raise()
|   8652 [garbagecat10] yeah, that's what I'm thinking. Now I have to try it.
+ 8674 [matz@ru y- a] I think you've solved this one.
  8683 [ruby-core@wh] Yes, I'll be fine.  I do have a small, related question about exception handling, matz.
  8695 [matz@ru y- a] ? It is assuming that argv[0] to be an instance of subclass of Module,
  8698 [mental@ry ia] A sandbox replaces even rb_cClass.  So, an exception raised by

[#5203] date.rb: DateTime.sec_fraction and day_fraction_to_time return incorrect second fraction
8562 [ruby-core@es] ...
+ 8563 [drbrain@se m] If you submit a patch with those changes I can check it in.
+ 8566 [sroberts@un ] I haven't been following this thread, don't know what the APIs are

[ ruby-Bugs-5361 ] accessing help from freeride
8567 [noreply@ru y] Bugs item #5361, was opened at 2006-08-09 02:02
8575 [drbrain@se m] On Aug 9, 2006, at 2:03 AM, <noreply@rubyforge.org>

[PATCH] Pathname.to_a
8568 [haisenko@co ] ...
+ 8571 [ara.t.howard] ...
| 8581 [haisenko@co ] Interesting, thanks.
+ 8580 [haisenko@co ] You're right, [-2] may be too simple but I think splitting the path into an
  8582 [hgs@dm .a .u] I've seen a subsequent reply to mine talking about Pathname#dirname