marc_marc4j reader producing ruby-marc recofds

gistfile1.txt

Testing on an 18K file in both marc21 and marc-xml. Loop looks like:

    reader = MARC::Reader.new(m21file) # or whatever appropriate reader
    reader.each do |r|
      t = r['245']['a']
    end

MARC version is the just-released 0.4.4

The following numbers are for a run with just enough compatibility to run the above code. 
It's very possible that with more indirection and more complex getters/setters things will
get a lot slower (i.e., this is mostly just measuring the speed to parse the records
and, in the case of marc/marc4j, stick the fields into a ruby-marc MARC::Record object).

                                   user     system      total        real
MRI 1.8.7 stock strict            51.600000   0.500000  52.100000 ( 52.097472)
MRI 1.8.7 stock forgiving         28.660000   0.320000  28.980000 ( 28.976728)
MRI 1.8.7 stock xml (nokogiri)    66.460000   0.430000  66.890000 ( 66.903439)
MRI 1.8.7 stock xml (libxml)      48.710000   0.370000  49.080000 ( 49.074222)

MRI 1.9.3 stock strict            19.780000   0.210000  19.990000 ( 19.997342)
MRI 1.9.3 stock forgiving         13.430000   0.070000  13.500000 ( 13.502137)
MRI 1.9.3 stock xml (nokogiri)    49.470000   0.360000  49.830000 ( 49.839829)
MRI 1.9.3 stock xml (libxml)      32.930000   0.260000  33.190000 ( 33.285764)

JRuby 1.9.2 stock strict          14.664000   0.000000  14.664000 ( 14.664000)
JRuby 1.9.2 marc4j strict          7.676000   0.000000   7.676000 (  7.676000)
JRuby 1.9.2 stock forgiving        9.543000   0.000000   9.543000 (  9.543000)
JRuby 1.9.2 marc4j forgiving      10.391000   0.000000  10.391000 ( 10.391000)
JRuby 1.9.2 stock xml (nokogiri)  43.023000   0.000000  43.023000 ( 43.023000)
JRuby 1.9.2 marc4j xml             4.893000   0.000000   4.893000 (  4.893000)