ruby-debug 0.10.2 → 0.10.3

data/CHANGES

@@ -1,3 +1,16 @@
+0.10.3
+11/17/08
+
+ - a backtrace now warns when it thinks the callstack is truncated which it
+   gets by comparing with caller()
+ - fix setting $0.
+ - fix bug in showing variables in post-mortem
+ - Document how Debugger.start with a block is intended to be used.
+ - Move Kernel#debugger from ruby-debug-base into ruby-debug
+ - Get regression tests working again
+ - Warn and add a "confirmation" when setting a breakpoint on a
+   file that is not loaded.
+
 0.10.2
  - debugger(steps=0) breaks inside of debugger rather than wait for a line event.
  - trace var varname (stop|nostop) added which issues trace_var.

data/ChangeLog

@@ -1,3 +1,310 @@
+2008-11-17 07:34  Rocky Bernstein
+
+	* CHANGES: Last commit before release.
+
+2008-11-16 00:14  Rocky Bernstein
+
+	* ChangeLog, bin/rdebug, lib/ChangeLog: Add rdoc for rdebug script.
+
+2008-11-15 22:13  Rocky Bernstein
+
+	* Rakefile, ext/ruby_debug.c: Go over RDOC documentation.
+
+2008-11-15 21:51  Rocky Bernstein
+
+	* doc/ruby-debug.texi: Add one more reference to start with a block
+
+2008-11-15 21:23  Rocky Bernstein
+
+	* doc/ruby-debug.texi: Remove hard-coded version number in example
+	  output
+
+2008-11-15 02:21  Rocky Bernstein
+
+	* ext/win32/.cvsignore: More administrivia
+
+2008-11-15 02:20  Rocky Bernstein
+
+	* ext/win32: More administrivia
+
+2008-11-15 02:18  Rocky Bernstein
+
+	* ., .cvsignore: More administrivia
+
+2008-11-14 19:39  Rocky Bernstein
+
+	* ., .cvsignore: More administrivia
+
+2008-11-14 19:38  Rocky Bernstein
+
+	* .cvsignore: Administrivia
+
+2008-11-14 19:37  Rocky Bernstein
+
+	* ., README, Rakefile, doc, doc/.cvsignore: Go over documentation
+	  for 0.10.3 release. rdoc creates files in doc/rdoc.
+
+2008-11-14 19:28  Rocky Bernstein
+
+	* README, configure.ac, doc/ruby-debug.texi,
+	  lib/ruby-debug-base.rb: Go over documentation and revise.
+
+2008-11-14 15:32  Rocky Bernstein
+
+	* CHANGES, ChangeLog, cli/ruby-debug.rb,
+	  cli/ruby-debug/commands/frame.rb, doc/ruby-debug.texi,
+	  ext/ruby_debug.c, lib/ChangeLog, lib/ruby-debug-base.rb: Move
+	  Debugger#debugger from base to cli. Revert code in ruby_debug.c
+	  and block parameter in debugger. cf. -> Compare with. Document
+	  Debugger.start better.
+
+2008-11-13 10:29  Rocky Bernstein
+
+	* ChangeLog, ext/ruby_debug.c, lib/ChangeLog: Make
+	  Debugger.start{block} work if Debugger.started? is false. Second
+	  try.
+
+2008-11-11 15:33  Rocky Bernstein
+
+	* ., Rakefile, test/cli/commands/catchpoint_test.rb,
+	  test/cli/commands/unit, test/cli/commands/unit/regexp.rb: Start
+	  unit test for command regular expressions. Much more willing out
+	  to be done later....
+
+2008-11-11 14:42  Rocky Bernstein
+
+	* test/data/raise.right: More line number changes in tdebug
+
+2008-11-11 14:40  Rocky Bernstein
+
+	* test/data/raise.right: tdebug lines have changed
+
+2008-11-11 02:07  Rocky Bernstein
+
+	* ChangeLog, cli/ruby-debug/commands/frame.rb, lib/ChangeLog,
+	  test/test-dollar-0.rb, test/test-frame.rb, test/trunc-call.rb:
+	  Tweak truncated stack test since Ruby's caller doesn't seem to
+	  include (tail?) recursive calls and we do. Get regression tests
+	  working in light of recent changes.
+
+2008-11-10 08:47  Kent Sibilev
+
+	* ext/ruby_debug.c: register debug_frame_t->arg_ary with GC
+
+2008-11-10 02:56  Kent Sibilev
+
+	* bin/rdebug, ext/ruby_debug.c: Trying to fix "if $0 == __FILE__"
+	  ruby's idiom. Apparently setting $0 to
+	  a new value doesn't work correctly. (Ruby's bug?)
+	  
+	  $ cat t3.rb
+	  p $0
+	  p File.expand_path $0
+	  $0 = File.expand_path $0
+	  p $0
+	  $ ruby t3.rb
+	  "t3.rb"
+	  "/Users/kent/Work/ruby-debug/trunk/t3.rb"
+	  "/Users/ke"
+
+2008-11-10 01:50  Kent Sibilev
+
+	* ext/ruby_debug.h: added declaration of
+	  check_breakpoints_by_method method to remove compiler warning
+
+2008-11-10 01:48  Kent Sibilev
+
+	* lib/ruby-debug-base.rb: a little bit more readable
+
+2008-11-10 01:43  Kent Sibilev
+
+	* ext/ruby_debug.c: running at_exit hooks at the end of the
+	  debug_load method in order to fix test cases a chance to run
+
+2008-11-10 01:41  Kent Sibilev
+
+	* ext/ruby_debug.c: can't store a copy of the debug_context in the
+	  stack like that, it doesn't play along with garbage collector.
+
+2008-11-10 01:39  Kent Sibilev
+
+	* ext/ruby_debug.c: these two methods are unused
+
+2008-11-10 01:36  Kent Sibilev
+
+	* ext/breakpoint.c: fix the compiler warning
+
+2008-11-10 01:35  Kent Sibilev
+
+	* lib/ruby-debug-base.rb: Debugger.start must always call the
+	  passed block
+
+2008-11-07 19:35  Rocky Bernstein
+
+	* ChangeLog, cli/ruby-debug/commands/frame.rb, lib/ChangeLog:
+	  Change truncated frame message.
+
+2008-11-07 10:39  Rocky Bernstein
+
+	* ChangeLog, bin/rdebug, cli/ruby-debug.rb,
+	  cli/ruby-debug/commands/frame.rb, lib/ChangeLog, test/tdebug.rb:
+	  Add check to "where" to see if the call stack is truncated; task
+	  #2354
+
+2008-11-06 16:17  Rocky Bernstein
+
+	* ChangeLog, Rakefile, lib/ChangeLog: #22698 Allow ruby-debug-base
+	  0.x.y.z be compatible with ruby-debug 0.x.y.
+
+2008-11-02 21:59  Rocky Bernstein
+
+	* ChangeLog, ext/ruby_debug.c, lib/ChangeLog,
+	  lib/ruby-debug-base.rb: Debugger.start with a block now stops
+	  inside the block. Debugger.debugger with a block works like
+	  Debugger.start with a block.
+	  
+	  The whole interface is hopelessly kludgy and needs to be redone.
+
+2008-10-27 15:29  Rocky Bernstein
+
+	* ChangeLog, test/data/raise.right: tdebug.rb line numbers changed.
+	  Update test numbers -- I think this is right.
+
+2008-10-26 14:54  Rocky Bernstein
+
+	* ChangeLog, ext/extconf.rb, ext/ruby_debug.c, lib/ChangeLog: Doc
+	  typo. Add comment to remind me how to turn off optimizationin
+	  extconf.rb
+
+2008-10-25 18:10  Rocky Bernstein
+
+	* cli/ruby-debug/interface.rb: Fix mismatched interface.
+
+2008-10-25 16:01  Rocky Bernstein
+
+	* cli/ruby-debug/commands/breakpoints.rb, ext/ruby_debug.c,
+	  lib/ChangeLog, test/data/annotate.right,
+	  test/data/breakpoints.right, test/data/emacs_basic.right,
+	  test/data/finish.right, test/test-finish.rb: Warn and add a
+	  "confirmation" when setting a breakpoint on a file that is not
+	  loaded. Regression tests no longer fail.
+
+2008-09-22 21:01  Rocky Bernstein
+
+	* cli/ruby-debug.rb: Remove a $DEBUG warning
+
+2008-09-22 00:07  Rocky Bernstein
+
+	* doc, doc/.cvsignore, lib/ruby-debug-base.rb,
+	  test/data/pm-bug.cmd, test/data/pm-bug.right,
+	  test/data/post-mortem.right, test/pm-bug.rb, test/test-pm.rb:
+	  #22118 bug in showing variables post mortem. Patch thanks to
+	  rubikitch.
+	  Update pm.rb integration test.
+
+2008-09-21 23:09  Rocky Bernstein
+
+	* CHANGES: In 0.10.3 territory now.
+
+2008-09-21 23:06  Rocky Bernstein
+
+	* ext/ruby_debug.c, test/tdebug.rb: tdebug.rb: modify frozen string
+	  taken from rdebug. Patch from Martin Krauskopf.
+	  ruby_debug.c: in 0.10.3 now.
+
+2008-09-03 17:33  Rocky Bernstein
+
+	* doc: Administrivia
+
+2008-09-03 17:29  Rocky Bernstein
+
+	* ChangeLog, lib/ChangeLog, test/helper.rb: Show line numbers when
+	  $DEBUG is set. Patch #21772 from Martin Krauskopf
+
+2008-08-28 05:43  Kent Sibilev
+
+	* ext/ruby_debug.c:
+
+2008-08-28 05:41  Kent Sibilev
+
+	* CHANGES: preparing version 0.10.2
+
+2008-08-15 17:36  Kent Sibilev
+
+	* cli/ruby-debug/commands/list.rb: Fixed an exception when calling
+	  Kernel#debugger method from the eval code
+
+2008-07-10 09:47  Kent Sibilev
+
+	* cli/ruby-debug/commands/continue.rb, cli/ruby-debug/processor.rb:
+	  allow 'continue' command from post-mortem
+	  
+	  here is an example where it is needed:
+	  
+	  ...
+	  Debugger.post_mortem do
+	  method_that_raises_an_exception()
+	  end
+	  ...
+	  
+	  I want to be able to inspect the post_mortem context and then to
+	  continue the program execution
+
+2008-07-10 08:55  Kent Sibilev
+
+	* ext/ruby_debug.c: fixed segfault and finish command
+
+2008-07-09 19:43  Kent Sibilev
+
+	* cli/ruby-debug/commands/breakpoints.rb: allow setting a
+	  breakpoint on a file that is not yet loaded
+
+2008-07-09 19:29  Kent Sibilev
+
+	* cli/ruby-debug/interface.rb: RemoteInterface should extend
+	  Interface class
+
+2008-07-09 19:25  Kent Sibilev
+
+	* bin/rdebug: $0 is frozen in Ruby 1.8.7
+
+2008-07-09 00:43  Rocky Bernstein
+
+	* cli/ruby-debug.rb, cli/ruby-debug/processor.rb: Some meager (and
+	  not complete) support for a verbose mode on
+	  source'ing a file. Some more print's turned to errmsg as
+	  appropriate.
+
+2008-07-07 07:11  Rocky Bernstein
+
+	* lib/ruby-debug-base.rb: Tracker [#20041] start erroneously moved
+	  to Kernel - should be in
+	  Debugger.start
+
+2008-07-03 16:03  Rocky Bernstein
+
+	* doc/ruby-debug.texi: Document that backslash can escape debugger
+	  command separation.
+
+2008-06-20 06:46  Rocky Bernstein
+
+	* CHANGES, cli/ruby-debug/commands/trace.rb,
+	  lib/ruby-debug-base.rb: trace.rb: add "trace var"
+	  ruby-debug-base.rb: remove another undefined warning.
+
+2008-05-27 03:54  Rocky Bernstein
+
+	* cli/ruby-debug/commands/save.rb, cli/ruby-debug/interface.rb,
+	  test/data/save.cmd, test/data/save.right, test/test-save.rb: Add
+	  save test. save.rb: use pretty-printed regexp. Interface, for now
+	  add restart_file accessor like the others. The processor
+	  interface
+	  need revision though.
+
+2008-05-24 01:27  Rocky Bernstein
+
+	* ChangeLog, doc/ruby-debug.texi, lib/ChangeLog: Remove dup lines.
+
 2008-05-23 14:57  Rocky Bernstein
 
 	* cli/ruby-debug/commands/save.rb: Catchpoints have changed. Fix

data/README

@@ -3,7 +3,8 @@
 == Overview
 
 ruby-debug is a fast implementation of the standard debugger debug.rb.
-The faster execution speed is achieved by utilizing a new hook Ruby C API.
+The faster execution speed is achieved by utilizing a new hook in the
+Ruby C API.
 
 == Requirements
 
@@ -28,20 +29,19 @@ ruby-debug is provided as a RubyGem.  To install:
 
 This should also pull in <tt>ruby-debug-base</tt> as a dependency.
 
-(If you install ruby-debug-base explicitly, you can add in the secret
---test option after "install" to have the regression test run before
+(If you install ruby-debug-base explicitly, you can add in the <tt>--test</tt>
+option after "install" to have the regression test run before
 installing.)
 
 For Emacs support and the Reference Manual, get
 <tt>ruby-debug-extra</tt>.  This is not a RubyGem, you'll need a Make
 program and a POSIX shell. With this installed, run:
 
-<pre>
   sh ./configure
   make
   make test # optional, but a good idea
   sudo make install
-</pre>
+
 
 ==== Install on MS Windows
 
@@ -86,13 +86,21 @@ code execution at run time.
     ...
   end
 
+or
+
+  require 'ruby-debug' ; 
+  Debugger.start do 
+    ...
+    debugger
+  end
+
 When Kernel#debugger method is executed, the debugger is activated
 and you will be able to inspect and step through your code.
 
 == Performance
 
-The debug.rb script that comes with the standard Ruby library uses
-Kernel#set_trace_func API. Implementing the debugger in pure Ruby has
+The <tt>debug.rb</tt> script that comes with the standard Ruby library uses
+<tt>Kernel#set_trace_func</tt> API. Implementing the debugger in pure Ruby has
 a negative impact on the speed of your program execution. This is
 because the Ruby interpreter creates a Binding object each trace call,
 even though it is not being used most of the time. ruby-debug moves

data/Rakefile

@@ -20,7 +20,10 @@ COMMON_FILES = FileList[
   'Rakefile',
 ]                        
 
-CLI_TEST_FILE_LIST = FileList['test/test-*.rb', 'test/cli/**/*_test.rb']
+CLI_TEST_FILE_LIST = FileList['test/cli/commands/unit/*.rb',
+                              'test/cli/commands/*_test.rb', 
+                              'test/cli/**/*_test.rb', 
+                              'test/test-*.rb'] 
 CLI_FILES = COMMON_FILES + FileList[
   "cli/**/*",
   'ChangeLog',
@@ -147,7 +150,7 @@ EOF
   spec.date = Time.now
   spec.rubyforge_project = 'ruby-debug'
   spec.add_dependency('columnize', '>= 0.1')
-  spec.add_dependency('ruby-debug-base', RUBY_DEBUG_VERSION)
+  spec.add_dependency('ruby-debug-base', "~> #{RUBY_DEBUG_VERSION}.0")
   
   # FIXME: work out operational logistics for this
   # spec.test_files = FileList[CLI_TEST_FILE_LIST]
@@ -217,13 +220,14 @@ end
 # ---------  RDoc Documentation ------
 desc "Generate rdoc documentation"
 Rake::RDocTask.new("rdoc") do |rdoc|
-  rdoc.rdoc_dir = 'doc'
+  rdoc.rdoc_dir = 'doc/rdoc'
   rdoc.title    = "ruby-debug"
   # Show source inline with line numbers
   rdoc.options << "--inline-source" << "--line-numbers"
   # Make the readme file the start page for the generated html
   rdoc.options << '--main' << 'README'
   rdoc.rdoc_files.include('bin/**/*',
+                          'cli/ruby-debug/commands/*.rb',
                           'lib/**/*.rb',
                           'ext/**/ruby_debug.c',
                           'README',

data/bin/rdebug

@@ -1,5 +1,106 @@
 #!/usr/bin/env ruby
-# -*- Ruby -*-
+
+#=== Summary
+#
+#A command-line front-end to the Ruby debugger, <tt>ruby-debug</tt>, the
+#Fast Ruby Debugger.
+#
+#Command invocation:
+#  
+#   rdebug [options] [--] [script-options] ruby-script-to-debug
+#   rdebug [options] [script-options] [--client]
+#   rdebug [--version | --help]
+#
+#=== Options
+#
+#<tt>-A | --annotate</tt> <i>level</i>::
+#    Set gdb-style annotation to <i>level</i>, a number.  Additional
+#    information is output automatically when program state is
+#    changed. This can be used by front-ends such as GNU Emacs to post
+#    this updated information without having to poll for it.
+#
+#<tt>--client</tt>::
+#    Connect to a remote debugger. Used with another rdebug invocation
+#    using <tt>--server</tt>.  See also <tt>--host</tt> and
+#    <tt>--cport</tt> options
+#
+#<tt>--cport=</tt><i>port</i>::
+#    Use port <i>port</i> for access to debugger control.
+#
+#<tt>-d | --debug</tt>::
+#    Set $DEBUG true.
+#
+#<tt>--emacs</tt>::
+#    Activates full GNU Emacs mode. Is the equivalent of setting the
+#    options <tt>--emacs-basic --annotate=3 --no-stop --no-control
+#    --post-mortem</tt>.
+#
+#<tt>--emacs-basic</tt>::
+#     Activates GNU Emacs mode. Debugger prompts are prefaced with two
+#     octal 032 characters.
+#
+#<tt>-h | --host=</tt><i>host</i>::
+#     Use host name <i>host</i> for remote debugging.
+#
+#<tt>-I | --include</tt> <i>path</i>
+#     Add <i>path</i> to <tt>$LOAD_PATH</tt>
+#
+#<tt>-m | --post-mortem</tt>::
+#     Activate post-mortem mode.
+#
+#<tt>--no-control</tt>::
+#      Do not automatically start control thread.
+#
+#<tt>--no-stop</tt>::
+#      Do not stop when script is loaded.
+#
+#<tt>-p | --port=PORT</tt>::
+#      Host name used for remote debugging.
+#
+#<tt>-r | --require</tt><i>script</i>::
+#      Require the library, before executing your script.
+#
+#<tt>--script</tt> <i>file</i>::
+#      Run debugger script file <i>file</i>
+#
+#<tt>-x | --trace</tt>::
+#      Show lines before executing them.
+#
+#<tt>--no-quit</tt>::
+#      Do not quit when script terminates. Instead rerun the
+#      program.
+#
+#<tt>--version</tt>::
+#      Show the version number and exit.
+#
+#<tt>--verbose</tt>::
+#      Turn on verbose mode.
+#
+#<tt>--v</tt>::
+#      Print the version number, then turn on verbose mode if
+#      a script name is given. If no script name is given
+#     just exit after printing the version number.
+#
+#<tt>--nx</tt>::
+#      Don’t execute commands  found in any initialization
+#      files, e.g. <tt>.rdebugrc</tt>.
+#
+#<tt>--keep-frame-binding</tt>::
+#      Keep frame bindings.
+#
+#<tt>--script=</tt><i>file</i>::
+#      Name of the script file to run 
+#
+#<tt>-s | --server</tt>::
+#      Listen for remote connections. Another rdebug session
+#      accesses using the <tt>--client</tt>  option. See also the
+#      <tt>--host</tt>, <tt>--port</tt> and <tt>--cport</tt> options
+#
+#<tt>-w | --wait</tt>::
+#      Wait for a client connection; implies <tt>-s</tt> option.
+#
+#<tt>--help</tt>::
+#      Show invocation help and exit.
 
 require 'rubygems'
 require 'optparse'
@@ -16,34 +117,11 @@ def debug_program(options)
     exit $?.exitstatus 
   end
   print "\032\032starting\n" if Debugger.annotate and Debugger.annotate > 2
-  unless options.no_rewrite_program
-    # Set $0 so things like __FILE == $0 work.
-    # A more reliable way to do this is to put $0 = __FILE__ *after*
-    # loading the script to be debugged.  For this, adding a debug hook
-    # for the first time and then switching to the debug hook that's
-    # normally used would be helpful. Doing this would also help other
-    # first-time initializations such as reloading debugger state
-    # after a restart. 
 
-    # However This is just a little more than I want to take on right
-    # now, so I think I'll stick with the slightly hacky approach.
-    $RDEBUG_0 = $0
+  # Record where we are we can know if the call stack has been
+  # truncated or not.
+  Debugger.start_sentinal=caller(0)[1]
 
-    # cygwin does some sort of funky truncation on $0 ./abcdef => ./ab
-    # probably something to do with 3-letter extension truncation.
-    # The hacky workaround is to do slice assignment. Ugh.
-    d0 = if '.' == File.dirname(Debugger::PROG_SCRIPT) and
-             Debugger::PROG_SCRIPT[0..0] != '.'
-           File.join('.', Debugger::PROG_SCRIPT)
-         else
-           Debugger::PROG_SCRIPT
-         end
-    if $0.frozen?
-      $0 = d0
-    else
-      $0[0..-1] = d0
-    end
-  end
   bt = Debugger.debug_load(Debugger::PROG_SCRIPT, options.stop, false)
   if bt
     if options.post_mortem
@@ -267,7 +345,7 @@ else
   # save script name
   prog_script = ARGV.shift
   prog_script = whence_file(prog_script) unless File.exist?(prog_script)
-  Debugger::PROG_SCRIPT = prog_script
+  Debugger::PROG_SCRIPT = File.expand_path prog_script
   
   # install interruption handler
   trap('INT') { Debugger.interrupt_last }