2215-2907 subjects 2697-2915

^ Nested popens under OSX
2467 [dave pragpro] - - - - - - - - - -
2512 [chad chadfow] I wasn't able to find the problem, but I can confirm that I see the

^ Change to #new (was OpenStruct#initialize to yield self)
2468 [dave pragpro] << patch chopped >>
+ 2469 [austin halos] How would it be overridden?
| 2470 [angus quovad] def initialize ...
| 2471 [austin halos] Ugh. I'm not sure that I like that.
| 2472 [dave pragpro] I'm not sure you'd notice any difference. If your constructors used a
| 2475 [austin halos] class Bar < Foo
| 2477 [dave pragpro] It _could_.... but whether it should is another matter :)
| 2479 [austin halos] I think that what I'm uncomfortable with is the possible "magic" involved
| 2481 [nobu.nokada ] It passes empty block to super, or you can pass no block by
+ 2473 [nobu.nokada ] It's an old fashion.
  2474 [dave pragpro] I know that in the case of File.new it is deprecated. File could be one

^ Exception not caught / require BUG
2491 [mailinglists] I found that the following exception does not raise an exception
2492 [decoux moulo] normal, LoadError don't inherit from StandardError

^ Re: [ruby-cvs] ruby: * dir.c (fnmatch): File.fnmatch with FNM_PATHNAME was broken
2493 [gsinclair so] The commit described below, affecting HEAD, appears to be a bug fix.
2505 [ocean m2.ccs] So, 1.8 branch is safe. (I forgot to check File.fnmatch with FNM_PATHNAME flag.)

^ rehash segfault
2494 [nathaniel ta] I don't have a lot of information on this bug at this point, but
2495 [matz ruby-la] Ruby hashes are not completely thread-safe.  I knew the problem
+ 2496 [nathaniel ta] Yes I can. Thanks for the definitive answer on the problem.
| 2500 [matz ruby-la] Unfortunately, detecting failure is the hardest part of the problem.
+ 2497 [vjoel PATH.B] Does this mean that some of the Hash methods are not atomic w.r.t. the
+ 2498 [eivind FreeB] What is the exact problem that makes it hard to fix without slowdown?
  2499 [matz ruby-la] Hash calls back a few Ruby methods, e.g. "hash", "eql?", etc, during
  2501 [eivind FreeB] This is a fun problem! :-)
  2502 [matz ruby-la] I'm glad you said so.

^ Thread-safe hashes (was Re: rehash segfault)
2503 [eivind FreeB] I believe full iterations would have to be sequenced alongside rehash;

^ foldl and foldr
2504 [ser germane-] charset="utf-8"
+ 2506 [eivind FreeB] foldl is available under the name inject.
| 2508 [ser germane-] charset="iso-8859-1"
| 2509 [eivind FreeB] Sorry - I read your original mail a little too fast.  Is this something
| 2510 [ser germane-] charset="iso-8859-1"
+ 2507 [matju sympat] for that naming back in 2000, as I later changed my mind.

^ [PATCH] trivial RDoc patch for array.c
2511 [djkea2 mugca] ===================================================================

^ $-<char> variables
2513 [dave pragpro] I couldn't find this in the Changelogs, so I thought I'd just ask.
2515 [matz ruby-la] $-<char> variable correspond with -<char> option to the interpreter.
2517 [dave pragpro] The reason I ask is that older interpreters used to allow  $-<anything>
2518 [matz ruby-la] This would be changed to be consistent.  Thank you for noticing it.
2519 [dave pragpro] and :$-) was _such_ a nice, happy symbol...

^ Possible bug in io.c
2514 [dave pragpro] static void
2516 [matz ruby-la] rb_define_hooked_variable("$stdout", &rb_stdout, 0, stdout_setter);

^ Bugs in source: nightly-snapshot
2520 [hgs dmu.ac.u] grabbed the nightly snapshot
2527 [dave pragpro] Thanks for the report - fixed.
2528 [hgs dmu.ac.u] OK, thanks, I'll have to refresh my understanding of all this, I was

^ Intelligent eyes needed
2521 [dave pragpro] I'm in the middle of updating the Pickaxe, and I wanted to try to
+ 2522 [dblack wobbl] (It may just be my eyes, but I think I see my last name in the first
+ 2523 [hgs dmu.ac.u] "The block may start with an argument list between vertical

^ STDIN.pos
2524 [dblack wobbl] $ /usr/local/src/ruby-1.6.8/ruby -ve 'STDIN.pos'
2525 [nobu.nokada ] $ ruby-1.6 -ve 'p STDIN.pos'
2526 [dblack wobbl] Actually, as I think about it, with STDIN I think pos doesn't matter.

^ concerns about Proc,lambda,block
2529 [dblack wobbl] Dave's post about the block/proc/Proc section in the Pickaxe got me
+ 2530 [hgs dmu.ac.u] I agree, I think this is something I'll have to keep looking up.
+ 2531 [decoux moulo] Well, the problem is perhaps that he try to rationalize (he must be
| 2533 [dblack wobbl] But there isn't any visual information to go on, once these objects
| + 2534 [austin halos] I think that what's more bothersome to me is that there doesn't appear to be
| | 2535 [dblack wobbl] I agree with you in principle, though I would rather see Proc and
| | 2536 [austin halos] We're on the same page, really. If they're going to be the same class, they
| + 2537 [matz ruby-la] I think these arity values should be different.  I will fix.
|   + 2540 [austin halos] Matz,
|   | 2546 [matz ruby-la] proc to lambda
|   | 2552 [jim freeze.o] From Dave's writeup I was under the impression that proc and
|   | 2553 [dave pragpro] If I gave that impression, it was in error: 'proc' and 'lambda' are the
|   | 2556 [matz ruby-la] No.  Where can I read your "writeup"?
|   | 2558 [dave pragpro] Cheers
|   | 2560 [matz ruby-la] I checked.  Some of your description is based on bugs I fixed already.
|   | 2562 [dave pragpro] That's strange, as the code in the book is formatted live with the
|   | 2565 [matz ruby-la] Perhaps I forgot to back-port the fix to 1.8.  The basic principle is
|   + 2542 [matju sympat] as few differences as it makes sense, because it's a mess explaining it to
|     + 2543 [austin halos] I agree with your suggestion for proc/method/block, because they're all
|     | 2554 [matju sympat] Well, I'm not of those who believe Ruby is not a multiple-inheritance
|     + 2545 [matz ruby-la] I hope so too.  The point is no one behavior can satisfy all.
|       2550 [batsman.geo ] To put it bluntly, we'll save some time if you answer to the following
|       + 2551 [dblack wobbl] A third possibility: keep them separate, but make them *more* like
|       + 2561 [matz ruby-la] My current stand point is that conceptually lambda returns a Proc
|         + 2563 [dave pragpro] I'm sorry to be stupid about this: normally I have no trouble
|         | 2567 [matz ruby-la] Usually when it happens, it is my fault.
|         | 2569 [eivind FreeB] While you're fixing this, will you also be changing so
|         | 2571 [matz ruby-la] I haven't decided yet.  Why don't you submit RCR?
|         | 2572 [eivind FreeB] I've submitted an RCR with a hopefully clear description, and an
|         + 2568 [hgs dmu.ac.u] I'm not sure what the right thing to do would be: sometimes
|           2570 [matz ruby-la] I don't know.  I wonder why you care if wrapped or not, where you
|           2573 [hgs dmu.ac.u] Because it affects how return and errors are handled. The method
|           2581 [matz ruby-la] The difference still lies in the Proc code body, which you can't read
|           2604 [austin halos] Will this restriction always be true?
|           2605 [matz ruby-la] I think so.  I'm not positive about disclosing AST or internal byte
+ 2532 [ser germane-] charset="iso-8859-1"

^ test/scanf failure
2538 [nobu.nokada ] test/scanf/test_scanfio.rb fails at `make test-all', and

^ Loading DLL in Ruby
2539 [v_kanetkar d] This is a multi-part message in MIME format.

^ Segfault in 1.9.0 tests
2541 [uehli bluewi] I get a segfault while running the ruby testsuite in the latest 1.9.0 CVS versions.
2544 [nobu.nokada ] We're trying to fix it.  (cf. thread from [ruby-dev:23038])
2555 [uehli bluewi] I just got a fresh cvs checkout and also deleted my whole

^ Accessing DLL from Ruby
2547 [v_kanetkar d] This is a multi-part message in MIME format.
2548 [bobcalco all] This is a multi-part message in MIME format.
2549 [v_kanetkar d] charset="iso-8859-1"

^ warning in test (Re: Segfault in 1.9.0 tests)
2557 [nobu.nokada ] Not pp.rb itself, due to test/inlinetest.rb.  Since older one

^ getoptlong.rb improvements
2559 [ehughes blue] The exception "AmbiguousOption" is misspelled, "quit mode" should read

^ Re: questions... about Proc,lambda,block
2564 [dave pragpro] I'm not sure I understand the orphaning rules. For example, in
2566 [matz ruby-la] m2 should not raise the LocalJumpError.  Now it is clear that I need
2583 [matz ruby-la] I've noticed I haven't explained well about the Proc orphan.

^ Comment football being played... with lib/test/unit.rb
2574 [nathaniel ta] ...

^ Comment football being played... with lib/test/unit.rb
2575 [nathaniel ta] [Resent because I accidentally signed it the first time]
2576 [dave pragpro] Oops - I missed the original change here.
2579 [gsinclair so] I agree.  My general philosophy is to put all the documentation at the
2580 [nathaniel ta] Sounds good. So long as we're all on the same page :-)

^ problem with Net::HTTP in 1.8.1
2577 [ian caliban.] #!/usr/bin/ruby -w
2578 [nobu.nokada ] HTTPResponse#to_ary returns an array contains itself, and #puts
2589 [ian caliban.] The patch fixes the recursive error, but I still can't get access to the
2591 [aamine lover] Use HTTPResponse#[] to access header fields.
+ 2592 [nobu.nokada ] Then, it should be warned?
+ 2593 [ian caliban.] Thanks for the help.

^ One more proc question
2582 [dave pragpro] Sorry about this... :)
2584 [matz ruby-la] You cannot tell the difference from a block from block arguments and a
2585 [jim freeze.o] def fred
2586 [matz ruby-la] Because I want to deprecate Proc.new without attached block eventually
+ 2587 [jim freeze.o] Currently I have some code that uses Proc.new. It was
| 2590 [matz ruby-la] Maybe, show me the code example.
+ 2608 [gsinclair so] I wouldn't worry about 1.8 -> 2.0 compatibility.  1.8 is good enough
  + 2609 [hgs dmu.ac.u] Though it would be nice if there were tools to help with this sort
  + 2613 [matz ruby-la] Although I said incompatibility is OK for 2.0, I demand a good reason

^ Duck typing chapter
2588 [dave pragpro] I've posted a rough first pass at a chapter about duck typing (and
+ 2595 [chad chadfow] # I've posted a rough first pass at a chapter about duck typing (and
| 2597 [dave pragpro] If you want to look at it again, I've updated it in light of your
| 2602 [chad chadfow] Thanks, Dave.  I think the tone is much less confrontational now.
+ 2596 [wkb airmail.] Dave,
  2598 [dave pragpro] Well, I think I'd probably say that we're interested in the class here,
  2619 [matju sympat] How are decimal and binary numbers not representations?
  2627 [ehughes blue] The second draft of the first few paragraphs is better than the first, but
  2629 [dave pragpro] I'm trying not to make it an argument. What I'm hoping to do is to lay
  2630 [ehughes blue] Those who haven't read the article will at this point assume that Stroustrup
  2637 [jm transact.] I tend to take the view of horses for courses,
  2639 [jm transact.] Re-read what I posted and realised I never said what I set out to say.

^ Net::FTP API change suggestion
2594 [djberge qwes] It seems to me that some redundancy in the Net::FTP module can be

^ Possible strangeness in Numeric
2599 [dave pragpro] To my surprise, Numeric defaults to implementing to_int by calling
2600 [matz ruby-la] This has been changed several times.  Some feel just like you, whereas
2601 [matju sympat] class Numeric
2603 [matz ruby-la] Very interesting idea.  I'm not sure whether this is what people (who

^ Thought about class definitions
2606 [dave pragpro] If we allowed
2607 [matz ruby-la] I can't explain the semantics for normal case, for example, in

^ process.c changes
2610 [djberge qwes] All,
2612 [nobu.nokada ] You may want to just extend Process::Status class?

^ process.c diff (oops)
2611 [djberge qwes] Here's the proper diff.
+ 2614 [matz ruby-la] I think it's better to turn off automatic text folding before posing
| + 2615 [djberge qwes] Sorry.
| | 2616 [matz ruby-la] We can name them 'wait3/wait4' or 'wait3/waitpid3' or something more
| | 2617 [djberg96 yah] They *are* called wait3 and waitpid3!  Now I'm really
| | 2618 [matz ruby-la] Did I say something confusing?
| + 2626 [akr m17n.org] wait3 is defined in SUSv2 but marked as legacy.
+ 2625 [akr m17n.org] ProcStruct is combination of pid, status and struct rusage.

^ bus error in eval.c
2620 [ryand zenspi] I found a bus error while debugging a pathological case between 1.6 and
+ 2621 [ryand zenspi] diff -d -u -r1.616.2.14 eval.c
+ 2622 [matz ruby-la] Seems fine.  Can you provide ChangeLog entry too?
  2623 [ryand zenspi] * eval.c (eval): Protected eval from backtraces that raise exceptions
  2624 [ryand zenspi] I'm not positive I have perms for that directory, but if so, I can

^ YAML complaint while generating RDoc
2628 [dave pragpro] With the latest CVS, I get
+ 2631 [dave pragpro] To answer my own question - no. It seems to be a problem with YAML. It
| + 2632 [ruby-core wh] Currently testing this.  Thanks.
| | 2641 [dave pragpro] Something had installed a syck shared library in site_ruby back in
| + 2633 [ruby-core wh] I'm unable to replicate this.  I'm going to borrow a friend's iBook and
|   2634 [dave pragpro] Today's the first time I've tried a complete build for a week or so.
+ 2635 [usa garbagec] Try make clean before make.
  2636 [dave pragpro] I did a 'make clean' and it seemed to make no difference. However,
  2638 [ruby-core wh] Not at all.  If it's any consolation, I just created a completely

^ patch to tempfile.rb to handle ENAMETOOLONG
2640 [vjoel PATH.B] Can anyone think of a better way to handle ENAMETOOLONG on systems that
2642 [matz ruby-la] Handling ENAMETOOLONG sounds nice, but is it OK to chop off path name,
2643 [vjoel PATH.B] Is the name of a tempfile significant, as long as it is unique? It looks
2645 [matz ruby-la] OK, I understand the whole problem now.  Thank you for clarify
2651 [nobu.nokada ] What about this?
2654 [matz ruby-la] Yep.  But I feel like this should be done easier.

^ RDoc proporsal
2644 [ocean m2.ccs] I found 'FIXME' in dir.c's RDoc.
2648 [dave pragpro] That's a good idea. But as I started to implement it, it struct me that
2653 [ocean m2.ccs] This is good enough. I'll replace 'FIXME' with '<i></i>'.
2655 [ocean m2.ccs] Sorry, it didn't work...
2656 [dave pragpro] You're right - I didn't think of that.
2657 [ocean m2.ccs] Don't worry, I'm not hurry.

^ Problems rdoc'ing cvs...
2646 [hgs dmu.ac.u] I have just done
2647 [dave pragpro] I'll look into wht test_syndication breaks RDoc, but in the meantime,
+ 2649 [gsinclair so] can browse the test coverage.  Examining the source code of someone
| 2650 [dave pragpro] Have you seen Hugh's problem: I can't reproduce it here?
| + 2658 [hgs dmu.ac.u] I went to a new directory, and re-got the rdoc cvs with the commands
| + 2666 [hgs dmu.ac.u] I've tried this again with a two new directories, one for rdoc and
|   2667 [dave pragpro] charset=US-ASCII;
|   2668 [hgs dmu.ac.u] OK, but I got the latest rdoc from CVS, and installed that before
|   2669 [dave pragpro] charset=US-ASCII;
|   2670 [hgs dmu.ac.u] Oh, OK.  Would it be possible to modify
+ 2652 [hgs dmu.ac.u] Thank you.

^ little glitch in String#split error reporting
2659 [dblack wobbl] Note that if you try to split on a Symbol, you get a slightly odd
2660 [matz ruby-la] Thank you!

^ Pathological slowdown in 1.8
2661 [ryand zenspi] I have some software that brings out a pathological condition that
2662 [matz ruby-la] I think the attached patch will fix the problem.  Thank you for
+ 2663 [ryand zenspi] That works perfectly. Thank you.
+ 3284 [ryand zenspi] Can I get this merged into the 1.8 branch, or is it frozen?
  3290 [matz ruby-la] Merged.  Thank you for reminding me.

^ DRb on Windows: the should-be-well-known problem with 'gets'
2664 [uehli bluewi] I just spent a day finding out why my DRb connections from a linux box

^ Re: Ruby Vim Download Link Broken?
2665 [shugo ruby-l] My ruby support was merged into the upstream version of vim, so

^ Rename #exception
2671 [dave pragpro] While I've been documenting all the to_int, to_str, methods, it
2672 [matz ruby-la] Not really the same purpose.
2673 [dave pragpro] Yes - the killer is the fact that the class version needs a parameter,

^ [PATCH] IO#fsync on fds opened as RO
2674 [batsman.geo ] Patch
2675 [matz ruby-la] Applied, this is a VERY good example of persuading me.

^ Infinite loop in st.c (st_foreach)
2696 [batsman.geo ] ruby 1.8.1 (2003-12-25) [i686-linux]