[Quantlib-dev] [ quantlib-Feature Requests-2568491 ] Coding standard errors in Quantlib

[SourceForge.net <no...@so...>]

Feature Requests item #2568491, was opened at 2009-02-05 15:27
Message generated for change (Settings changed) made by lballabio
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=2568491&group_id=12740

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Priority: 5
Private: No
Submitted By: Marcel Stampfer (marcelstampfer)
Assigned to: Nobody/Anonymous (nobody)
Summary: Coding standard errors in Quantlib

Initial Comment:
Hi Luigi,

I'm just getting to know Quantlib and thought that a way I could get started with writing some code would be to do some static analysis. Doing this, I've found some potential bugs in the code that could be easily patched. Some of the errors that I've found have to do with the follow violations:

Calling virtual functions from constructors and destructors
Using local variables that have the same name as class/parent class variables 
Not initializing all member variables in constructors
Downcasting a pointer to an abstract base class to a subclass
Making implicit conversions from one type to another
Passing objects by reference instead of by value (more an optimization issue)
Not throwing an exception by value or catching by reference

Do you think that any of these errors warrant patching? I could give concrete examples if you like but should download the SVN head beforehand as I'm using the release code. 

Regards
Marcel

----------------------------------------------------------------------

Comment By: Marcel Stampfer (marcelstampfer)
Date: 2009-02-06 13:47

Message:
Okay. My plan is to run the examples and unit tests through a debugger to
see if any of these violations are reached (there are 100's). With that I
can compare patched versions to determine if any of the violations are bugs
in the code.

Marcel

----------------------------------------------------------------------

Comment By: Luigi Ballabio (lballabio)
Date: 2009-02-06 10:53

Message:
Yes, please go ahead.

Luigi


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=2568491&group_id=12740