Dear sir
what spicificly I want to measure are to levels of Quality , first
is the design level , I will give you our main list of targets to measure
(It can be expanded later) :
* Detect and report cyclic inheritance (this should be cought on compile
time but not by all compilers or all languages)
* calculate depth of inheritance (number of parents) { we can't say when the
depth of inheritance has exceeded proper level but the user can decide that
from the report we generate }
* Detect and report the number of children of a class.
* calculate the Weighted Methods per Class (WMC)
* Detect and report high cyclomatic complexity within methods. {Also there
is no defined number for HIGH but the user can define it in run time or the
tool will generate a report of weights of classes and he chooses what to say
to be HIGH }
* Detect methods implemented as one big loop or heavy weighted loops.
* Detect and report Intermodule dependency. {Class coupling , Inheritance
and aggregation, etc..}
* Detect and report modules that can be candidates to be reusable modules.
* Detect and report the Using message passing as primary inter-process
communication. { decrease efficiency at runtime }
That's our main design issues that we plan to detect and report till now if
the base parsing engine and analyser is built this list can be expanded to
include other flaws.
on the other side there is the code level , we intended to ensure C++
checklist provided by OpenOffice.org but we didn;t know it is out of date ,
we need to find a decent alternative in ADDITION to it , our Java specific
checklist will be gained from Sun developers community , we will generate a
generic checklist that could be used for any OOP language .
code conventions supported will be those of OpenOffice.org mainly plus we
will use GNU http://www.gnu.org/prep/standards/standards.html code standars
(we will not be able to deliver all)
Now to your Question, yes I've checked what tools already exist in this area
commercial and open source , they all concentrate on code level , few
detected design issues (.NET code analysis tool was a great deal) but my
tool will be better/different for the following reasons I hope :
* This tool will be mainly designed for openoffice.org standards so it will
be more usable for OOo developers.
* This tool will be free and open source.
*This tool will include more than one language and will enable language
insertion at run time intelligently (major difference in our plan)
* This tool will be able to focus on design and code analysis in the same
time without affecting efficiency of any , any new issues will be plugged
in easily.
* Generating XMI files of design generated from code will enable many design
tools (mainly ARGOUML) to read it and view digrams of the code.
That's our main goal of improving our work than the current existing tools ,
as soon as we generate any prototypes I will inform you for your comments
(we have now an XMI parser and analyser for only four of the design issues
mensioned).
Thanks for your time and support
waiting for your comments.
Mohammed.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
