I was thinking of using AbstractLinkedList as it might save the need to have dedicated TreeList iterator classes, ie. the ones from AbstractLinkedList could be used instead. What is more important is to get the Fadelung code into the CVS version before release 3.1 ;-)
Stephen ----- Original Message ----- From: "J�rg Schm�cker" <[EMAIL PROTECTED]> > I am sorry, that I did not do the F�delung-code on the tidied up version. I > must admit I am not familiar with the AbstractLinkedList contract. I just > checked it. What would be the advantage of doing so? It does not extend > the interface/contract just helps implementing it. Am I missing something? > > Joerg > > > -----Original Message----- > From: Stephen Colebourne [mailto:[EMAIL PROTECTED] > Sent: Monday, May 10, 2004 10:44 PM > To: J�rg Schm�cker > Cc: Jakarta Commons Developers List > Subject: Re: [collections] TreeList > > > I have checked in the version of TreeList that I had been working on (tidied > up from your original code). The F�delung code looks interesting, so should > probably replace the checked in code. Also, I think that it might be > possible to have TreeList subclass AbstractLinkedList, which would be nice. > > Stephen > > ----- Original Message ----- > From: "J�rg Schm�cker" <[EMAIL PROTECTED]> > > Stephen, > > > > I have added a fast implementation for iterator() and listIterator() using > a > > thing called "F�delung" in German. Please copy the Apache license into the > > files. > > > > Thanks, > > > > Joerg > > > > -----Original Message----- > > From: Stephen Colebourne [mailto:[EMAIL PROTECTED] > > Sent: Wednesday, May 05, 2004 1:46 AM > > To: J�rg Schm�cker > > Cc: Jakarta Commons Developers List > > Subject: Re: [collections] TreeList > > > > > > The default iterator is picked up from AbstractList. For each next() it > uses > > get(index) on the list to retrieve the element. As you might imagine, > > ArrayList can easily beat this with a dedicated implementation going > > straight to the array. > > > > The two possible solutions are to have each node in the tree refer to its > > parent, or to have the iterator dynamically use an ArrayStack to build up > > the parent structure as it navigates it. I favour the latter if possible. > > > > Stephen > > > > ----- Original Message ----- > > From: "J�rg Schm�cker" <[EMAIL PROTECTED]> > > > Can you tell me how the iterator is implemented? Currently, I have not > > > implemented such forward/backward linking to improve iteration > > performance. > > > > > > Regards, > > > > > > Joerg > > > > > > -----Original Message----- > > > From: Stephen Colebourne [mailto:[EMAIL PROTECTED] > > > Sent: Monday, May 03, 2004 4:00 PM > > > To: Jakarta Commons Developers List > > > Cc: J�rg Schm�cker > > > Subject: Re: [collections] TreeList > > > > > > > > > Thanks. Forwarded to commons-dev for reference. > > > > > > I will be checking the class in shortly. It does need a proper iterator > > > implementation, as the default is slow. This can be added later though. > > > > > > Stephen > > > > > > ----- Original Message ----- > > > From: "J�rg Schm�cker" <[EMAIL PROTECTED]> > > > > Sorry for the hassle. I am very happy to contribute the code. Next > > time > > > I > > > > will contribute with the standard Apache licence. > > > > > > > > I am hoping that I'll have the time to finish the tree-based class to > > > > implement a ordered subset of a set witht the same order and fast > > > insertion. > > > > That was the original reason I started implementing it as a tree. The > > idea > > > > that it will be much faster came to me only during the development. > > Don't > > > > know if it is a standard algorithm, but I did not copy it anywhere. > > > > AVL-Trees should not be a problem. > > > > > > > > Regards, > > > > > > > > Joerg > > > > > > > > -----Original Message----- > > > > From: Stephen Colebourne [mailto:[EMAIL PROTECTED] > > > > Sent: Sunday, May 02, 2004 11:34 AM > > > > To: J�rg Schm�cker > > > > Subject: Re: [collections] TreeList > > > > > > > > > > > > You're essentially stating that you haven't blatently ripped the code > > off > > > > from somewhere else (except standard computer science algorithm > books). > > > Also > > > > that, if it was written for a company, that the company doesn't claim > > any > > > > ownership. > > > > > > > > IANAL, but my understanding is that legally you are donating a COPY of > > > your > > > > code to Apache that will then be OWNED by Apache. Both you and Apache > > can > > > > then do as they please with the code - in Apache's case release it > with > > > > commons collections. > > > > > > > > Effectively, you are saying that you COULD sign > > > > http://www.apache.org/licenses/#clas without having to go through the > > > hassle > > > > of doing so. > > > > > > > > BTW, I am only having to mail you this because you didn't submit your > > > > original code with an Apache licence. If you had, I would have taken > > that > > > as > > > > the sign of donation! > > > > > > > > Stephen > > > > > > > > ----- Original Message ----- > > > > From: "J�rg Schm�cker" <[EMAIL PROTECTED]> > > > > > Stephen, > > > > > > > > > > Could you give me a reason, why I should not be in the position to > > > donate > > > > > the class? I don't think there is one but I just want to make sure. > > > > > > > > > > Joerg > > > > > > > > > > -----Original Message----- > > > > > From: Stephen Colebourne [mailto:[EMAIL PROTECTED] > > > > > Sent: Saturday, May 01, 2004 11:57 PM > > > > > To: [EMAIL PROTECTED] > > > > > Cc: Jakarta Commons Developers List > > > > > Subject: [collections] TreeList > > > > > > > > > > > > > > > Joerg, > > > > > I have been looking at integrating your tree based list into commons > > > > > collections. Can you confirm that you are in a position to donate > the > > > > class > > > > > to Apache please? A reply to confirm this will allow its inclusion. > > > > > > > > > > Stephen > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
