CERN-BARC Video Conference

Wednesday 11 Oct 2006, 10:30 → 11:00 Europe/Zurich

CERN

Regular CERN-BARC Video conference on ELFms developments BARC phone: +91 22 25595228

Paper xml_parsing_comparision.pdf

Sharma: Dear German, Here is my status report Wassh2: New version of wassh with all the decided modifications has been installed on lxdev12 and available for testing. Discussed the work related to Notification system with German and Thorsten. The design document is being prepared.. regards Sharma -------------------------- Murthy: Hi German, My status report is as below, 1. CVS commits done for encryption configuration setup (task 1 of the total work) 2. There is an RFE from Miro for TCP cache management modification. This is required to prevent surge on lemon server upon it's coming online after shutdown. In the new scheme disk cache is removed and a limited amount of memory cache is maintained in a circular buffer. The changes are almost done and testing is planned now. regards murthy --------------------------- Salim: Hi German, Here is my status report:- 1. Extended XML-Gateway to handle comma separated list of node-names...each node-name entry may be an individual node-name or a PHP regexp pattern.(Done for both OraMon as well as FlatMon). 2. Extended Lemon API to support add_node() and remove_node() methods. Testing is done for both OraMon as well as FlatMon. 3. Conducted extensive study on XML parsing techniques available in libxml2. Out of the three major techniques(DOM,SAX, & XmlTextReader API) I found that XmlTextReader API has edge over other two techniques:- * DOM results in a document loaded completely in memory (The size of the document that can be handled is limited by the size of memory available) * SAX works by registering callbacks as it progresses through the XML document. This Programming model is relatively complex, not well standardized, can't provide validation directly, makes entity, namespace & base processing relatively hard. * XmlTextReader API provides far simpler programming model, has direct support for namespaces, base and entity handling, & adding DTD validation on top of it is relatively simple. * XmlTextReader API provides a simpler, more standard & more extensible interface to handle large documents than the existing SAX version. * Both SAX and XmlTextReader API does not load whole document into memory. * XmlTextReader API is closely modelled after XmlTextReader & XmlReader classes of the C# language, and is assumed to be a replacement of SAX. I'm testing the parsing performance by writing code for parsing our XML document in XmlTextReader API. Moreover it seems to me that implementing node and metric Iterator is far easy in XmlTextReader API. I'm attaching a comparision chart between major XML parsing tecniques (I found this on Net). Regards, Salim --------------------------------- Dinesh: Dear German / Bill , Last week I had nothing new to report on CCTracker except setup & testing (at BARC) all the development done at CERN during my stay over there. Now I'm working on implementing new feature - delete & rename of object & rack as communicated by Bill and then prepare the new release of CCTracker. Best regards, Dinesh ------------------------------------ Discussion Miro-Salim: Hi Miro, The XmlTextReader support is added to libxml2 in version 2.5.0. and in SLC3 the default RPM version is 2.5.10....I think we have no problem working with libxml2/XmlTextReader API on SLC3 platform. Cheers, Salim Miroslav Siket wrote: > Hello Salim, > > I have discussed the issue of libxml2 with linux support and it looks > like SLC3/RHES3 will not come with 2.6+ version and that SLC3 will be > discontinued Sep 2007. We have a solution though - you can go against > libxml2 with XMLReader as defined in libxml2, but we need to > statically build the XML API. > > Cheers, > Miro