1836-2120 subjects 2215-2907

[PATCH] STARTTLS support for Net::SMTP
2057 [will@gl ze .] ...

RDoc problem
2058 [gsinclair@so] # I'm not sure where to report RDoc bugs at the moment since the
2063 [dave@pr gp o] You can always e-mail me :)
2065 [dave@pr gp o] Oops - now I see what you're saying: the class is documented, but the

[PATCH] Rhapsody configure.in bug
2060 [sunshine@su ] Here is a patch which fixes a bug in Ruby's configure.in which causes
2061 [eban@os ri .] Thanks.  Applied.

I'm being stupid, I know, but my yaml's broken in 1.9
2069 [dave@pr gp o] ruby 1.9.0 (2003-12-25) [powerpc-darwin]
2070 [dave@pr gp o] OK - I found (I think) the problem. I build out of the box, then

[patch] rdoc is broken in 1.8.1
2071 [a.bokovoy@sa] Greetings!
2072 [dave@pr gp o] Cheers
2073 [nahi@ke na t] You commited it to main trunc but it's for 1.9 already.
2074 [dave@pr gp o] OK - In general I haven't been updating the changelog for minor stuff.
+ 2075 [matz@ru y- a] It's upto maintainers.  You decide.
+ 2076 [nahi@ke na t] As matz said, it's up to you.
  2079 [dave@pr gp o] My thinking was that I'm in the middle of a work-in-progress, basically
  2094 [nahi@ke na t] It must be "I just didn't think ...".
  2097 [dave@pr gp o] The main change I'm making is altering every .c file in the
  + 2100 [gsinclair@so] Unlikely; a lot of stuff gets added for test/unit, ext/tk, and others,
  + 2107 [nahi@ke na t] I have always believed that it's just the interpreter.  All bundled

2077 [decoux@mo lo] Why svg is stupid ?
2078 [akr@m1 n. rg] Because {IO,StringIO,Zlib::GzipReader}#read(nil) is modified to return

Is this troubling?
2080 [dave@pr gp o] ./configure
2081 [decoux@mo lo] What is this line ?
2083 [ehughes@bl e] I think it's the nanosleep test. I get a complaint about ./configure line
2098 [decoux@mo lo] Well, on a normal system
2099 [ehughes@bl e] yeah, now i'm awake i was going to take a look, but it's already been fixed

Using the ruby_1_8 branch
2082 [gsinclair@so] NaHi made a comment recently (in ruby-core:2073) about committing to
2085 [matz@ru y- a] CVS HEAD is now for 1.9.
2091 [gsinclair@so] If I take the (b) option, does somebody have to back port it for every
2095 [nahi@ke na t] It was snow in Japan.  How was your hot Xmas?

Error with Socket.getaddrinfo on OS X
2084 [rich@in oe h] ruby 1.8.1 (2003-12-25) [powerpc-darwin]
+ 2086 [rich@in oe h] irb(main):002:0> require 'socket'
| 2088 [matz@ru y- a] Maybe --with-lookup-order-hack=INET would help.
| 2090 [rich@in oe h] Makes no difference.
+ 2087 [matz@ru y- a] I have another report that getaddrinfo(3) is not working properly on
| 2089 [rich@in oe h] File socket/mkmf.log
+ 2092 [rich@in oe h] Same problem posted in RubyTalk # 84909
  2093 [rich@in oe h] MacOS X Panther has a lousy getnameinfo() implementation that

no error in getaddrinfo...
2096 [rich@in oe h] Socket.do_not_reverse_lookup=true

[BUG] Can't call to_s on a frozen Date
2101 [gsinclair@so] Interesting...
2108 [matz@ru y- a] Because Date object cache calculated values.
2109 [gsinclair@so] Sorry about that; I didn't mean for my reply-to to be set like that.
2110 [matz@ru y- a] I understand.  The point is how we can fix, or not.
2115 [g@in er ec -] It may add a slight overhead, but what about changing the once method?

[BUG] syck segfaults when used in rdoc
2102 [a.bokovoy@sa] Greetings!
2103 [decoux@mo lo] yes, see my message on ruby-talk. The bug is in syck_emitter_write() it
2104 [a.bokovoy@sa] Yes, I stated same (look at my previous email, closer to its end). The problem is
2105 [decoux@mo lo] * it add 9362 (len) to marker
+ 2106 [decoux@mo lo] it's (marker - buffer), sorry
+ 2111 [matz@ru y- a] @@ -234,3 +234,3 @@ syck_emitter_write( SyckEmitter *e, char
  2112 [decoux@mo lo] Well, I don't understand sorry. For me the problem is in the for (;;)
  2113 [matz@ru y- a] Like this?
  + 2114 [decoux@mo lo] yes,
  + 2116 [a.bokovoy@sa] This patch fixes the problem. Thank you!

Strange paths on Ruby build under Windows
2117 [dave@pr gp o] I'm trying to track down an RDoc problem under Windows, but I'm coming
2118 [chad@ch df w] #
2119 [dave@pr gp o] Thank you, Chad - that was it.

Interesting webrick errors
2121 [nathaniel@ta] With my application running for a long time under heavy load, I've

Bad interaction between timeout.rb and --enable-pthread
2122 [nathaniel@ta] require 'test/unit'

sleep is broken with --enable-pthread [Was: Bad interaction between timeout.rb and --enable-pthread]
2123 [nathaniel@ta] I should have reduced it more before posting...
+ 2124 [chad@ch df w] Looks like the select() call in rb_thread_wait_for is being interrupted
| 2125 [decoux@mo lo] Well, if I've understood (with this new year I'm still stupid :-))
| 2129 [chad@ch df w] I wish I could be Ruby-stupid just like you, Guy ;)
+ 2131 [matz@ru y- a] Nathaniel, can you try this?
  + 2132 [decoux@mo lo] small typo, I think
  + 2133 [matz@ru y- a] I left small typo in the patch.  Try this one instead.
    2134 [nathaniel@ta] It passes the test with flying colors, and fixes the unit tests for my
    2135 [nathaniel@ta] /usr/lib/ruby/1.8/webrick/httprequest.rb:307: [BUG] rb_sys_fail() -
    + 2136 [matz@ru y- a] Can you extract the code to reproduce the problem?
    + 2137 [matz@ru y- a] This patch might work.
      2138 [nathaniel@ta] I'll give it a go... I'm trying to reduce the problem to something

2126 [decoux@mo lo] pigeon% ./configure --enable-pthread
2127 [matz@ru y- a] CVS latest is
2128 [decoux@mo lo] Yes, I've looked in CVS before posting, the problem is here
2130 [matz@ru y- a] Yes, I should have noticed that at the first time.  I just fixed it.

Best way to install ri documentation
2139 [dave@pr gp o] All the RDoc documentation for the core interpreter is now complete.
+ 2140 [gsinclair@so] Congratulations.
| 2141 [matz@ru y- a] It may be convenient for users without source code, but I worry about
| 2145 [rich@in oe h] Perhaps make it available for mirrors and save ruby-lang's bandwidth?
| + 2146 [gsinclair@so] [Rich, please avoid top-posting.]
| + 2147 [dave@pr gp o] I'd really like it just to be there (and that was a big part of the
| + 2151 [ruby@ja es r] We can mirror the complete Ruby source and the ri/rdoc bundle on
+ 2142 [jlst@ge ty g] My 2 cents is that anything that can be derived from source should be
  2143 [dave@pr gp o] This sounds reasonable to me: rdoc documentation gets built during
  2144 [jlst@ge ty g] For sure.  I was thinking

Re: Best way to install ri documentation -- please check this
2148 [dave@pr gp o] So, I'm thinking about doing the following? Is this OK with everyone?
2149 [jlst@ge ty g] Does this mean that you have to have root in order to 'make rdoc'?
+ 2150 [gsinclair@so] Yes, but you have to be root to do 'make install' anyway.  Or do you?
+ 2152 [dave@pr gp o] splitting the target into a root and non-root part. Given that it seems
  + 2153 [nobu.nokada@] Is DESTDIR effective to --ri-system?
  | 2154 [dave@pr gp o] base    = File.join(Config::CONFIG['datadir'], "ri", version)
  | 2219 [jfh@ci e. fl] The first problem with using DESTDIR I've run into is that ruby may not
  | + 2222 [jfh@ci e. fl] Generating RI...
  | | 2226 [dave@pr gp o] I still need to understand all this DESTDIR business. When I need to
  | | 2227 [eivind@Fr eB] DESTDIR is a prefix used for installing ONLY.  It's sort of a chroot for
  | + 2223 [jfh@ci e. fl] One other thing...do you want to use dependency targets for the docs
  | + 2224 [dave@pr gp o] Matz is concerned about the bandwidth overhead of this.
  |   2228 [jfh@ci e. fl] Ok...FWIW, we (cise.ufl.edu) would be glad to be a ruby mirror for any
  |   2230 [dave@pr gp o] What effect does DESTDIR have on ~, the user's home directory. It's
  |   + 2232 [jfh@ci e. fl] Hmmm...it should have no effect on a homedir...it really should have no
  |   + 2233 [eivind@Fr eB] DESTDIR is not relevant for ~/.rdoc, really - it is used only during
  |     2234 [dave@pr gp o] I've checked in a patch, but I have no idea if it does what is needed.
  |     2241 [jlst@ge ty g] Here's 2 more cents.  The output dir used by rdoc should be set via a
  |     2242 [dave@pr gp o] Does the existing patch work, though...
  |     + 2243 [jlst@ge ty g] gmake[1]: Leaving directory `/usr/home/jfh/dev/extern/ruby/ext/zlib'
  |     + 2250 [jfh@ci e. fl] It seems to work fine, with the exception that it still requires
  |       2252 [dave@pr gp o] Does "make test" work in that environment, because testrunner.rb does
  |       2254 [nahi@ke na t] See [ruby-core:02168].
  |       + 2259 [dave@pr gp o] The surprising thing to me is that I thought YAML could fall back on
  |       | + 2261 [jfh@ci e. fl] LD_LIBRARY_PATH=$(DESTDIR)/usr/local/lib $(DESTDIR)/usr/local/bin/ruby -I\
  |       | | 2262 [dave@pr gp o] When you say "installs ri" do you mean the command, or the
  |       | | 2264 [jfh@ci e. fl] Sorry, the ri docs.
  |       | + 2263 [nahi@ke na t] yaml/parser is deprecated. [ruby-talk:72859]
  |       + 2265 [jlst@ge ty g] The problem I encountered yesterday,
  + 2155 [eivind@Fr eB] Adding an extra minute to all 'make install's sounds like it will hinder
  | 2156 [dave@pr gp o] Do people do a 'make install' while hacking? For most stuff I just run
  | + 2157 [matz@ru y- a] I don't.
  | + 2161 [nahi@ke na t] Some bundled library developers(*1) must do it for test (test/runner.rb).
  |   2162 [dave@pr gp o] The last thing I want to do is to make anyone unhappy. I'm more than
  |   + 2164 [jlst@ge ty g] What about using autoconf?
  |   | 2165 [dave@pr gp o] That'd be cool, but I'm clueless about autoconf. If folks feel this is
  |   + 2166 [nahi@ke na t] I'm just reporting.  Don't worry too much.  Once agreed, developers
  |     2167 [dave@pr gp o] Could I use the same trick that 'make test' does, and simply add the
  |     + 2168 [nahi@ke na t] +1 to keep simple.  We Japanese developers once abandoned to run
  |     | + 2169 [gsinclair@so] Philosophically, developers (the minority) should sacrifice some
  |     | | + 2171 [nahi@ke na t] Developers build ruby everyday (I build 2-3 times per day).  Some
  |     | | + 2182 [ruby@ja es r] What he said.
  |     | + 2170 [dooby@d1 .k ] Embrace change, NaHi :)
  |     |   2172 [nahi@ke na t] I always try to do so as Weinberg said. :)
  |     |   2174 [jim@fr ez .o] For CVS
  |     |   2178 [gsinclair@so] Forgive my ignorance, but how does one effect this?
  |     |   2179 [jim@fr ez .o] I don't know if it is the right thing to do, but I thought
  |     + 2173 [jlst@ge ty g] How about making it disabled by default for now and to consider making
  |     + 2175 [jlst@ge ty g] Here's a quick attempt at providing autoconf support for building rdoc
  |       2176 [matz@ru y- a] Thank you.  Looks nice.  Try to unfold the patch next time.
  |       2177 [dave@pr gp o] Looks great (although nahi said we'd need to use full Ruby, not
  |       2180 [jlst@ge ty g] How about this instead?
  |       2181 [nahi@ke na t] Agreed.  bin and doc (if user wants) should be built and installed
  |       2183 [dave@pr gp o] Would install be dependent on rdoc?
  |       2184 [nahi@ke na t] Sorry.  I had not to mix the topic.  According to the topic about
  |       2185 [jlst@ge ty g] This was my first reaction as well.  Unfortunately, rdoc needs to
  + 2158 [jlst@ge ty g] First of all, let me apologize for being naturally paranoid.
    2159 [dave@pr gp o] Good point. I'll add an option to let you force 'ri' just to use a
    2160 [jlst@ge ty g] Sure would.  Thanks.

Occasional --enable-pthread hangs...
2163 [nathaniel@ta] First of all, thanks so much to all those that have helped with
2235 [nathaniel@ta] My process has just hung again at this point, and this time attaching
2236 [decoux@mo lo] Well, the problem with such error (waiting in wait_for_tcp_memory()) is
2237 [nathaniel@ta] OK... any good way of determining if it is a kernel problem?
2238 [decoux@mo lo] give an example (real script) to try to reproduce your problem. Sorry to
2239 [nathaniel@ta] No need to be sorry... I know they're not much to go on. This latest

Absolute paths in shebang lines?
2186 [jlst@ge ty g] What would the reaction be to reconsidering the following proposal?
2187 [matz@ru y- a] It's OK if there's reasonable reason.
2188 [jlst@ge ty g] Sorry I was a little vague.  Here's what I did that caught me up.
2190 [matz@ru y- a] I'm not sure I understand you correctly.  The installation process
2192 [jlst@ge ty g] Err.  I think I'm the one confused.  Never mind.

[Patch] lib/test/unit/collector/dir.rb
2189 [nahi@ke na t] ...
2191 [nathaniel@ta] So you've tested the patch against all of test/ and it only breaks DRb?
2193 [nahi@ke na t] Yes.  But I installed all ext libs as I possible.

[PATCH] File.readable_world? and File.writable_world?
2194 [ian@ca ib n.] ...
+ 2195 [eivind@Fr eB] (1) There is no such thing as "readable world" in the permissions of
| 2199 [ian@ca ib n.] World-readable _is_ a term applied very often in the UNIX world to the
| 2200 [matz@ru y- a] I prefer world_writable? that means being writable from anybody.
| + 2201 [ian@ca ib n.] Do you want me to submit a new patch?
| | 2202 [matz@ru y- a] Yes, hopefully with ChangeLog entry.
| | 2204 [ian@ca ib n.] ...
| | + 2205 [nobu.nokada@] I'll feel happy if File::Stat also has them.
| | + 2207 [dave@pr gp o] I just wanted to say a quick "thank you" for including the RDoc
| + 2203 [ian@ca ib n.] What about writable_others? I think that would be useful to have, too.
|   2206 [matz@ru y- a] I think it would be source of confusion.  They are too similar.  How
|   2208 [ian@ca ib n.] That's an interesting idea. The only problem is that world_writable?
|   2209 [matz@ru y- a] The mode 446 is -r--r--rw-, which is writable from others, so that
|   + 2210 [ian@ca ib n.] I'm sorry. I misread what you had written. I thought you wanted the mode
|   + 2214 [ian@ca ib n.] Here is a new patch that implements things the way you describe. I've
|   | 2220 [matz@ru y- a] Thank you.  It's the best way to contribute.  Merged.
|   | 2240 [ian@ca ib n.] The patch I sent you contained the lib/pathname.rb methods.
|   | 2246 [ian@ca ib n.] Hmm, I guess I somehow failed to include the patch to lib/pathname.rb.
|   | 2249 [ian@ca ib n.] OK. Here's the patch for lib/pathname.rb with ChangeLog.
|   + 2256 [ian@ca ib n.] ...
|     2271 [ian@ca ib n.] I still don't see this patch applied to CVS. Is it OK, Matz?
|     + 2272 [matz@ru y- a] Don't worry.  It's OK.  I just need to work on other problem before
|     | 2273 [ian@ca ib n.] OK, thanks.
|     | 2276 [matz@ru y- a] No.  I will fix this.
|     + 2274 [nobu.nokada@] BTW, I think this cast for BCC wouldn't be needed at all.  Even
|       2275 [ian@ca ib n.] Good to know. I simply copied that from elsewhere in the code, because I
+ 2197 [sunshine@su ] To avoid complaints from GCC when warnings are enabled, these should be

[PATCH] NextStep/OpenStep defines
2196 [sunshine@su ] Here is a patch which ensures that the useful stat.h constants (S_IRUSR,
2198 [matz@ru y- a] Will be merged.  Thank you.

2211 [dave@pr gp o] I notivce that there're a bunch of new xxx_init_copy methods: RDoc is
2212 [nobu.nokada@] They are called interally by clone/dup like as initialize
2213 [dave@pr gp o] I don't suppose I could persuade the author to add RDoc comments
2312 [nobu.nokada@] How can I write __classpath__ literally?  RDoc never stop
2313 [dave@pr gp o] You can write \_\_classpath__ :   backslash stops interpretation of
2314 [nobu.nokada@] Thank you, why didn't I try two backslashes?
2315 [dave@pr gp o] Is this for cases where there isn't already a dup or a clone method? Or
2325 [nobu.nokada@] Now #dup and #clone belong to Kernel, and they call
2326 [dave@pr gp o] Does that mean that there's no difference between dup and clone for
2329 [matz@ru y- a] "clone" does
2330 [dave@pr gp o] Thanks, matz.
2331 [nobu.nokada@] Sorry for my bad explanation, they are different, of course.
2332 [dave@pr gp o] OK - I'm sorry I misunderstood. I'd love to help you out here: what
2333 [nobu.nokada@] Thank you, [ruby-core:02326] may be enough.  Or, more
2334 [dave@pr gp o] This is why I asked the question about dup and clone being the same.
2338 [matz@ru y- a] Difference lies in the Object#dup and Object#clone.  initialize_copy
2339 [dave@pr gp o] Ah - I finally see. Sorry to be so dense.