Subject: please answer
From: Karl Lieberherr (lieber@ccs.neu.edu)
Date: Fri Dec 07 2001 - 10:00:12 EST
Hi Doug:
please can you answer this question from Germany.
Is the AP Library now used by both DJ and DemeterJ?
Is it correct that the *.trv files only give an approximation of the
real traversal by collapsing the traversal graph.
We did this because in most cases the approximation is accurate.
But in any case, DemeterJ generates the correct traversal code.
To Sven:
http://www.ccs.neu.edu/research/demeter/biblio/new-strategy-semantics.html
is a follow-up paper to the paper you cite. It has
a simpler semantics.
-- Karl
>From ap-support-admin@lists.ccs.neu.edu Fri Dec 7 05:06:17 2001
>Date: Fri, 07 Dec 2001 11:07:18 +0100
>From: Sven Buergel <sven.buergel@informatik.tu-chemnitz.de>
>To: ap-support@lists.ccs.neu.edu
>Content-Transfer-Encoding: 8bit
>Subject: [Ap-support] traversal graph representation in DemeterJ
>X-BeenThere: ap-support@lists.ccs.neu.edu
>X-Mailman-Version: 2.0.4
>List-Help: <mailto:ap-support-request@lists.ccs.neu.edu?subject=help>
>List-Post: <mailto:ap-support@lists.ccs.neu.edu>
>List-Subscribe: <https://lists.ccs.neu.edu/bin/listinfo/ap-support>,
> <mailto:ap-support-request@lists.ccs.neu.edu?subject=subscribe>
>List-Id: <ap-support.lists.ccs.neu.edu>
>List-Unsubscribe: <https://lists.ccs.neu.edu/bin/listinfo/ap-support>,
> <mailto:ap-support-request@lists.ccs.neu.edu?subject=unsubscribe>
>
>Hello,
>
>after reading the strategies papers [1] I got a bit curious about the
>internal representation of traversal graphs (TG) in DemeterJ.
>
>In the paper a TG consist of a number of appropriately pruned
>class-graph copies, i.e. one copy per strategy-graph edge. However, if I
>look at the generated *.trv files of DemeterJ, I can only see one pruned
>class graph regardless of how many edges the corresponding strategy
>graph has. It seems that in this case all class subgraphs are merged
>into one graph.
>
>So, it would be interesting to me whether DemeterJ's internal TG
>representation really conforms to the one stated in the paper or the one
>given by *.trv files.
>
>Regards
>Sven Bürgel
>
>[1] Lieberherr/Patt-Shamir, 2000: Traversals of Object Structures:
>Specification and Efficient Implementation.
>
>_______________________________________________
>Ap-support mailing list
>Ap-support@lists.ccs.neu.edu
>https://lists.ccs.neu.edu/bin/listinfo/ap-support
>
This archive was generated by hypermail 2b28 : Fri Dec 07 2001 - 10:05:10 EST