Ruby » Ruby 1.8

I can't find a protect_against_forgery? in ruby 1.8, and action_view is rails code.  I think you meant to file this bug on dev.rubyonrails.org.

Hi Eric!

I don't get this error with
ruby 1.8.6 (2008-03-03 patchlevel 114) [i686-linux]
so it must be caused by
ruby 1.8.6 (2008-06-20 patchlevel 230) [i686-linux]
(By the changes since p114.)

It would be great if you could try to reproduce this error, perhaps line 21 is enough to trigger it.

You'll need to provide more information to reproduce this, as there is too little information and too much work required:

$ ./script/server
-bash: ./script/server: No such file or directory

So I'd need to install rails and build a rails app.  Can you reduce this to a reproduction without rails?

Duplication of #199 I believe. Fixed in the repository.

I compiled the current Ruby 1.8.6:

$ svn co http://svn.ruby-lang.org/repos/ruby/branches/ruby_1_8_6
[... make etc ...]
$ /home/tobi/bulk/run/ruby/ruby_1.8.6_rev_17790/bin/ruby -v
ruby 1.8.6 (2008-07-01 patchlevel 257) [i686-linux]
[... installed RubyGems, Rails, etc ...]
[In the Rails app's dir:]
$ ed script/server
97
1
#!/usr/bin/env ruby
1i
#!/home/tobi/bulk/run/ruby/ruby_1.8.6_rev_17790/bin/ruby
.
wq

Result:

$ ./script/server
=> Booting Mongrel (use 'script/server webrick' to force WEBrick)
=> Rails application starting on http://0.0.0.0:3000
[...]
# browsed to http://localhost:3000/
# -> no error.

$ ./script/server -e test -p 2000
[...]
# Browsed to index.
[...]
$ # No segfault.

$ ./script/server -e production --port 4000
# Browsed to index
[...]
$ # No error.

I can't talk about any related issues, but the issues I experienced
are fixed.

Thanks for the fixes!

In the future it would be great if the Ruby community would be asked
to test a release candidate before a real version of Ruby is released and
recommended as stable. (No matter if the release will be made in response
to a security issue or not.)

Thanks for Ruby!

I got this exact error too. Reproducing it was a bit awkward, but I think I can confirm the problem and that it has been fixed.
In my case, I was finding that in my replica live environment:

* "script/server" worked fine
* "mongrel_rails start -P tmp/pids/mongrel.pid" usually demonstrated the bug: "wrong argument type FalseClass (expected Proc)"
* "mongrel_rails" by itself caused a segfault:

 $ mongrel_rails start
 ** Starting Mongrel listening at 0.0.0.0:3000
 ** Starting Rails with development environment...
 ** Rails loaded.
 ** Loading any Rails specific GemPlugins
 ** Signals ready.  TERM => stop.  USR2 => restart.  INT => stop (no restart).
 ** Rails signals registered.  HUP => reload (without restart).  It might not work well.
 ** Mongrel 1.1.4 available at 0.0.0.0:3000
 ** Use CTRL-C to stop.
 Segmentation fault

And I've also seen:

 $ mongrel_rails start -P tmp/pids/mongrel.pid
 ** Starting Mongrel listening at 0.0.0.0:3000
 ** Starting Rails with development environment...
 ** Rails loaded.
 ** Loading any Rails specific GemPlugins
 ** Signals ready.  TERM => stop.  USR2 => restart.  INT => stop (no restart).
 ** Rails signals registered.  HUP => reload (without restart).  It might not work well.
 ** Mongrel 1.1.4 available at 0.0.0.0:3000
 ** Use CTRL-C to stop.
 *** glibc detected *** corrupted double-linked list: 0x00715858 ***
 Aborted

This is RHEL4 with p230 built from the Fedora package "ruby-1.8.6.230-2.fc10.src.rpm", which includes a few patches of its own, including "ruby-1.8.6.230-string-str_buf_cat.patch"

So to replicate this, on a test box (Ubuntu 6.06.1) I made two completely fresh vanilla installations compiled from svn, tags v1_8_6_230 and v1_8_6_264

 ---------------- how ruby was installed ------------
 svn co http://svn.ruby-lang.org/repos/ruby/tags/v1_8_6_230 tags/v1_8_6_230
 cd tags/v1_8_6_230
 autoconf
 ./configure --prefix=/usr/local/ruby-1.8.6p230
 make
 sudo make install
 wget 'http://rubyforge.org/frs/download.php/38646/rubygems-1.2.0.tgz'
 rm -rf rubygems-1.2.0
 tar -xvzf rubygems-1.2.0.tgz
 cd rubygems-1.2.0
 sudo /usr/local/ruby-1.8.6p230/bin/ruby setup.rb
 sudo /usr/local/ruby-1.8.6p230/bin/gem install rails --version "2.0.2" --no-rdoc --no-ri
 # Rails requires a database layer (is there a way to disable models?)
 sudo /usr/local/ruby-1.8.6p230/bin/gem install sqlite3-ruby --no-rdoc --no-ri
 ----------------------------------------------------
(repeat for p264)

 ---------------- sample rails app ----------------
 rails foo
 cd foo
 script/generate controller bar
 cat <<EOF >app/views/bar/index.html.erb
 <% form_tag(:action=>:index) do %>
 <% end %>
 EOF
 /usr/local/ruby-1.8.6p230/bin/ruby script/server
 # In another console window
 lynx http://localhost:3000/bar/
 --------------------------------------------------

I had to hit the server several times, but after 4 or 5 tries under p230 this caused a crash:

 $ /usr/local/ruby-1.8.6p230/bin/ruby script/server
 => Booting WEBrick...
 => Rails application started on http://0.0.0.0:3000
 => Ctrl-C to shutdown server; call with --help for options
 [2008-07-02 13:22:39] INFO  WEBrick 1.3.1
 [2008-07-02 13:22:39] INFO  ruby 1.8.6 (2008-06-20) [i686-linux]
 [2008-07-02 13:22:39] INFO  WEBrick::HTTPServer#start: pid=2634 port=3000
 127.0.0.1 - - [02/Jul/2008:13:22:41 BST] "GET /bar HTTP/1.1" 200 179 - -> /bar
 127.0.0.1 - - [02/Jul/2008:13:22:43 BST] "GET /bar HTTP/1.1" 200 179 - -> /bar
 127.0.0.1 - - [02/Jul/2008:13:22:43 BST] "GET /bar HTTP/1.1" 200 179 - -> /bar
 127.0.0.1 - - [02/Jul/2008:13:22:44 BST] "GET /bar HTTP/1.1" 200 179 - -> /bar
 *** glibc detected *** corrupted double-linked list: 0xb7f5b358 ***
 Aborted

With p264, I could not crash it (tried 30 hits)

Trying again with mongrel instead of webrick:

 sudo /usr/local/ruby-1.8.6p230/bin/gem install mongrel --no-rdoc --no-ri
 /usr/local/ruby-1.8.6p230/bin/ruby script/server    # option 1
 /usr/local/ruby-1.8.6p230/bin/mongrel_rails start   # option 2

After 4 or 5 hits, mongrel (1.1.5) under p230 crashed (run either as option 1 or option 2). But p264 didn't crash even after 40 hits.

In conclusion: whilst I haven't been able to replicate exactly the problem "wrong argument type String (expected Proc)" using a clean build, I could replicate crashes reliably, and these have gone away in the recent 1.8.6 tree.

Aside: in order to make a new RPM package for my live environment, it would be helpful if there could be a new 1.8.6 tarball release...