- From: Aleksey Sanin <aleksey@aleksey.com>
- Date: Fri, 15 Mar 2002 16:40:50 -0800
- To: TAMURA Kent <kent@trl.ibm.co.jp>, w3c-ietf-xmldsig@w3.org
FYI: I tested my C14N and Exc-C14N implementation (http://xmlsoft.org) using the same test file (exc-signature.xml in http://www.w3.org/Signature/2002/02/01-exc-c14n-interop.html). The test program loaded the document once and executed (exc-)c14n 10000 times on PIII-900 + Linux 2.4.18 (the time for loading and parsing document is *NOT* included) with following results: C14n: 4550 [ms] Exclusive C14n: 4500 [ms] There were no visible difference in doing c14n with or without comments. The program was compiled using gcc-2.96 using w/o optimization (debug mode). Joseph, please remove question marks for libxml on iterop report (http://www.w3.org/Signature/2002/02/01-exc-c14n-interop.html) since it does support "#default" prefix and the performance does exceed c14n performance. Thanks, Aleksey. TAMURA Kent wrote: >This is a report with an IBM implementation > http://www.alphaworks.ibm.com/forum/xmlsecuritysuite.nsf/current ># The latest public release does not have an exclusive c14n ># implemenataion conforming to CR yet. > >Not rendering ...: Y1 >The first occurence ..: Y1 >InclusiveNamespaces ..: Y1 >The "#default" ...: Y1 >Performance: > C14n: 53037 [ms] > Exclusive C14n: 72584 [ms] > Xerces2 serializer: 13219 [ms] > These are time for serializing a DOM tree representing a small > document 10000 times. Note that we have not done any > optimization yet. >
Received on Friday, 15 March 2002 19:42:14 UTC