Re: [Boost-users] [icl] possible bug for set difference

Subject: Re: [Boost-users] [icl] possible bug for set difference
From: Joachim Faulhaber (afojgo_at_[hidden])
Date: 2011-07-18 18:26:35

2011/7/18 Thomas Taylor <thomas.taylor_at_[hidden]>:
> Hi!
>
> First let me express my happiness for having this library, as it makes a lot
> of my programming tasks much easier!

Hi Thomas,

thank you for the nice feedback. I am always happy to hear from people
who like Boost.Icl :)

> While using it I think I discovered a bug for interval_set, when a
> difference is calculated; please find a test case attached. This was
> compiled against recent svn (r73190).

I've read your example. You are using an Interval-class with a "point
and length" representation which is interesting. I never did tests
using such intervals but I think everything should work correctly.

As you found out yourself, the default ctor Interval() has to produce
an empty Interval and yes, you have to take care for underflow, if
member Corner is an unsigned integral type.

I hope everything works well now and you have fun using icl.

Cheers,
Joachim

-- 
Interval Container Library [Boost.Icl]
http://www.joachim-faulhaber.de