Improving std.regex(p)

[Ellery Newcomer]

On 06/21/2010 04:07 AM, Ben Hanson wrote:
> == Quote from Ellery Newcomer (ellery-newcomer at utulsa.edu)'s article
>> On 06/19/2010 03:35 AM, Ben Hanson wrote:
>>> I can't prove what the comlexity is for DFA compilation, but instead, I
>>> challenge anyone to show slow compilation times with any DFA compatible regex.
>>> As I don't have some D code yet, you could try the current lexertl code (just
>>> add a single rule). Compile as Release if using Visual C++, as the iterator
>>> debugging kills performance in VC++ 8 or above (one problem I guess wouldn't
>>> exist with a D codebase).
>>>
>>> As I say, the only obvious area for me that hammers compilation times is huge
>>> repeats. If you repeat an expression 50,0000 times or something, then yes, you
>>> will get a huge machine and it will take time to construct! :-)
>> et unicode?
>>>
>>> There are papers on tackling this with counting repeats, even for DFAs. If this
>>> area interests anyone, please let me know.
>> it does
> Try this link: http://www.arl.wustl.edu/~mbecchi/files/becchi_ancs2009.pdf

thanks