Stealing eXpat's lunch

[Finnian Reilly]

Benchmarking Eiffel Xpact and C eXpath

USING: eXpat XML parser (pure C)
Parsing: libexpat/testdata/largefiles/ns_att_test.xml
Parsing time: 164 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <z:row> occurrences 50000
TAG: <rs:data> occurrences 1
TAG: <xml> occurrences 1

USING: Xpact XML parser (Eiffel)
Parsing: libexpat/testdata/largefiles/ns_att_test.xml
Parsing time: 133 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <z:row> occurrences 50000
TAG: <rs:data> occurrences 1
TAG: <xml> occurrences 1


USING: eXpat XML parser (pure C)
Parsing: libexpat/testdata/largefiles/recset.xml
Parsing time: 144 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <z:row> occurrences 50000
TAG: <s:AttributeType> occurrences 32
TAG: <s:datatype> occurrences 32
TAG: <rs:data> occurrences 1
TAG: <s:ElementType> occurrences 1
TAG: <s:Schema> occurrences 1
TAG: <s:extends> occurrences 1
TAG: <xml> occurrences 1

USING: Xpact XML parser (Eiffel)
Parsing: libexpat/testdata/largefiles/recset.xml
Parsing time: 124 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <z:row> occurrences 50000
TAG: <s:AttributeType> occurrences 32
TAG: <s:datatype> occurrences 32
TAG: <xml> occurrences 1
TAG: <s:Schema> occurrences 1
TAG: <s:ElementType> occurrences 1
TAG: <rs:data> occurrences 1
TAG: <s:extends> occurrences 1


USING: eXpat XML parser (pure C)
Parsing: libexpat/testdata/largefiles/wordnet_glossary-20010201.rdf
Parsing time: 70 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <rdf:Description> occurrences 99642
TAG: <b:glossaryEntry> occurrences 49964
TAG: <rdf:RDF> occurrences 1

USING: Xpact XML parser (Eiffel)
Parsing: libexpat/testdata/largefiles/wordnet_glossary-20010201.rdf
Parsing time: 65 ms
Tags sorted in order of occurrence count (Highest first)

TAG: <rdf:Description> occurrences 99642
TAG: <b:glossaryEntry> occurrences 49964
TAG: <rdf:RDF> occurrences 1

[Finnian Reilly]

The code: C and Eiffel

Bench mark script in example: run/*.sh

[Finnian Reilly]

Correcting an unfair advantage for eXpat tag counter

I just realized I unintentionally gave the expat tag counter an unfair advantage by compiling it as a single threaded application. After recompiling the Eiffel tag counter we now have a new set of benchmarks. (inlining set to 5)