Hi Michael,
We'd like to thanks you for taking time to respond. We chose to analyse
xerces because it is a high-quality library that has successfully evolved
to support many different standards.
The answers you provided supports our hypothesis that many "bad code"
detection techniques which measure some sort of complexity will
raise false alerts as they ignore important factors like the complexity of
the domain.
For your information, this is not a stand-alone investigation. We
have been listening in to this mailing list over one year. The list has
provided us information about the current development process and team
composition. We have also tracked bug reports and feature requests to
classes for every version of the system.
cheers,
Stephane Vaucher & Foutse Khomh
On Thu, 11 Jun 2009, Michael Glavassevich wrote:
Hi,
I don't normally respond to surveys (of any kind) but I will say that the
classes that you've listed represent solutions to large / complex problems
(like XML schema validation) and that none of them were quick and dirty.
Thanks.
Michael Glavassevich
XML Parser Development
IBM Toronto Lab
E-mail: [email protected]
E-mail: [email protected]
Stephane Vaucher <[email protected]> wrote on 06/10/2009 01:45:47
PM:
Dear Xerces developers,
In the course of a academic study of software quality, we built a
quality model
to identify problem areas (classes) in software systems and applied it to
Xerces. We relied on the judgement of students to evaluate how good the
detections are at identifying god classes [1]. We would like confirmation
by
some developers (or experienced users) of a subset of the classes
detected.
All that is required is to indicate :
1/ if indeed the class centralises functionality and is a god class;
2/ whether or not the class structure is coherent with a clear design
decision. More precisely, does the class offer a good solution to a
problem or
was it introduced as a quick and dirty solution which should eventually
be
corrected (given available resources, and given a need other than
elegance)?
Your participation will help us evaluate the usefulness of the modelfor
real
developers and subsequently improve the detection model. The classes
are listed
below, so you can just reply to this email with your comments/opinions.
* * * * * * * *
Class 1 : org.apache.xerces.impl.xpath.regex.RegularExpression
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 2 : org.apache.xerces.impl.dv.xs.XSSimpleTypeDecl
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 3 : org.apache.xerces.xinclude.XIncludeHandler
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 4 : org.apache.xerces.impl.xs.XMLSchemaValidator
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 5 : org.apache.xerces.parsers.AbstractDOMParser
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 6 : org.apache.xerces.impl.dtd.DTDGrammar
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 7 : org.apache.xerces.impl.XMLEntityManager
God class: yes/no
Design: good solution/quick and dirty
Comment:
Class 8 : org.apache.xerces.dom.DOMNormalizer
God class: yes/no
Design: good solution/quick and dirty
Comment:
* * * * * * * *
The model used to produce this list be published at the international
conference on quality software (QSIC2009) [2].
[1] classes that tend to centralise functionality
[2] http://home.ewha.ac.kr/~bjchoi/conference/QSIC2009/
Stephane Vaucher & Foutse Khomh
PhD students - Software quality
University of Montreal
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
