0

I have written a Ruby version of Erik Demaine's (MIT) docdist8.py. This is available on github as docdist-v3.rb. I faced two weird kind of situations:

1) In the function inner_product there is a block comment:

Inner product between two vectors, where vectors
are repeated as dictionaries of (word, freq) pairs.
Example : inner_product({"and":3, "of":2, "the":5},
                        {"and":4, "in":1, "of":1, "this":2}) = 14.0

If I wrap this with =begin and =end there is no problem, but if I wrap it with triple double-quotes """, I get errors as follows:

./docdist-v3.rb:71: syntax error, unexpected tIDENTIFIER, expecting kEND
    Example : inner_product({"and":3, "of":2, "the":5},
                                         ^
./docdist-v3.rb:71: syntax error, unexpected tIDENTIFIER, expecting kEND
    Example : inner_product({"and":3, "of":2, "the":5},
                                                  ^
./docdist-v3.rb:72: syntax error, unexpected kIN, expecting kEND
...                 {"and":4, "in":1, "of":1, "this":2}) = 14.0
                              ^
./docdist-v3.rb:72: syntax error, unexpected tIDENTIFIER, expecting kEND
...         {"and":4, "in":1, "of":1, "this":2}) = 14.0
                              ^
./docdist-v3.rb:72: syntax error, unexpected tIDENTIFIER, expecting kEND
..."and":4, "in":1, "of":1, "this":2}) = 14.0
                          ^

Are there rules / allowed entries for """ that are different from =begin and =end?

2) When I run my program with the time command it executes in about 0.3 seconds. However, if I put require 'profile' the time it takes becomes very high in comparison - 30 seconds. Hence I don't get the correct output at all. This doesn't seem to be the case with the original Python version, where it takes only a marginal extra time to profile. How do I get the same profile run in Ruby?

Note: The two files I used to run the Ruby program are t2.bobsey.txt and t3.lewis.txt. They are available at http://ocw.mit.edu/ans7870/6/6.006/s08/lecturenotes/dd_data.htm

Fabio Cardoso
  • 1,181
  • 1
  • 14
  • 37
user1575148
  • 561
  • 6
  • 28

1 Answers1

0

1) Block comments always have the form:

=begin
Comment
=end

You are actually creating a string that is never used:

"""
Not a comment
"""
# => "\nNot a comment\n"

That's why you get an error when adding another quote and that's why the syntax highlighting renders them as strings.

2) It's slower with profiler but I get identical results:

ruby docdist-v3.rb t2.bobsey.txt t3.lewis.txt
File t2.bobsey.txt:262111 lines,49785 words,3354 distinct words
File t3.lewis.txt:1015474 lines,182355 words,8530 distinct words
The distance between the documents is: 0.574160 (radians)
Stefan
  • 109,145
  • 14
  • 143
  • 218
  • Thanks for clarifying the """ mystery to me. I have changed all comment blocks to =begin and =end now in docdist-v5.rb [link](https://gist.github.com/mh-github/5379949) What times do you get without require 'profile' and with require 'profile'? For me, from a normal execution time of about 0.3 seconds, Ruby takes about 30.5 seconds with profile. It also does not report the execution time accurately. This is not the case with the Python original. Hence my question : how do we profile in Ruby without adding so much overhead and get cpu time accurately, without using the time command? – user1575148 Apr 13 '13 at 20:51
  • Maybe [ruby-prof](https://github.com/ruby-prof/ruby-prof) is a better alternative. BTW, you can just update your existing Gist, GitHub keeps a revision history. – Stefan Apr 13 '13 at 21:27
  • Thanks for the gist tip. I was using Ruby 1.8.6. ruby-prof requires 1.8.7 or higher. So I installed Ruby 1.8.7. Strangely docdist-v5.rb runs slower with Ruby 1.8.7. ruby-prof is reporting zero seconds time in its output. Looks like I'll have to stick to 1.8.6 and time command for the moment. Seems to me that the only way to speed up the program now is to Ruby to provide something similar to Python's string.maketrans utility which will be faster than tr. – user1575148 Apr 14 '13 at 05:32