[Mono-devel-list] System.Text.RegularExpression Benchmark

[Eric Durand-Tremblay]

Hi all,

Before begining my work on ILBased regex,  I composed a litle benchmark 
to compare the actual interpreted mono regex to MS regex in the two modes.

The grid below represent the times in millisecond to find all occurence 
of each regex in the HTML help of Mysl and glibc combined. (more or less 
10mo)

Regex                                                                  
                               | NbMatch   | Mono  | MS interpreted | MS 
Compiled
mysql                                                                  
                                         6934        235                 
       20                     20
(?<left><(?<inter>[a-zA-Z]+)>)[^<>]*</\k<inter>>                        
48810      2000                    2183                    631
(\s[ab]{2,3}\w+\s)                                                                                       
1810      2600                    2884                    912
\w{3,5}://([\.]|\w|\\)*                                                                                       
556      8250                    4286                  2193
[abC-DefgG-Jklm]{5}                                                                                 
1231      4100                    2454                    751
(\w|(\w\.\w)|(\w_\w)|(\w\-\w))+@(\w|(\w\.\w)|(\w_\w)|(\w\-\w))+\.\w{2,4}           
334    63020                  31906                15312

I attached the code of this benchmark if anybody's intrested to verify 
it's validity.

Note that the gain of performance with the compiled mode of MS is very 
intresting.

---------------------------
Eric Durand-Tremblay
University Laval
Québec (Canada)
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: RegexBench.cs
Url: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20040110/ac7c8fbf/attachment.pl